Lincoln University Digital Thesis Copyright Statement The digital copy of this thesis is protected by the Copyright Act 1994 (New Zealand). This thesis may be consulted by you, provided you comply with the provisions of the Act and the following conditions of use: 


you will use the copy only for the purposes of research or private study you will recognise the author's right to be identified as the author of the thesis and due acknowledgement will be made to the author where appropriate you will obtain the author's permission before publishing any material from the thesis. Partitioning biotic and abiotic components of soil CO2 fluxes using
subsurface CO2 dynamics and stable carbon isotopes,
Taylor Valley, Antarctica
A thesis
submitted in partial fulfilment
of the requirements for the Degree of
Doctor of Philosophy
at
Lincoln University
by
Fiona L. Shanhun
Lincoln University
2013
ii
Abstract of a thesis submitted in partial fulfilment of the
requirements for the Degree of Doctor of Philosophy
Abstract
Partitioning biotic and abiotic components of soil CO2 fluxes using subsurface
CO2 dynamics and stable carbon isotopes, Taylor Valley, Antarctica
by
Fiona L. Shanhun
To date, surface CO2 fluxes in soils of the McMurdo Dry Valleys in Antarctica have been assumed to
represent heterotrophic respiration. Important aspects of soil ecosystem function have been inferred
on the basis of this assumption.
This study used high-resolution measurements of CO2 concentration, and, for the first time, the
stable C isotopic composition of surface CO2 fluxes and subsurface CO2 profiles to: 1) link surface and
subsurface CO2 dynamics, 2) test whether abiotic mechanisms of CO2 production and consumption
can explain diel variability in surface CO2 flux rates, and 3) partition biotic and abiotic components of
surface CO2 fluxes at two sites with contrasting organic C levels in soils of Taylor Valley.
In the 2008/09 austral summer, surface CO2 flux rates and subsurface CO2 concentration and δ13CCO2
profiles were measured at the warmest and coolest times of the day at two sites with contrasting soil
organic C contents. Site A, located near a small lake, can be considered a “biological hotspot”, with
the lake providing a contemporary organic C source via algal growth and subsequent wind-dispersal
onto nearby soils. Site B had no contemporary lacustrine-derived organic C supply, and soil organic C
levels were typical of the low levels in Dry Valley soils. A physical model, describing the temperaturedriven dissolution and exsolution of CO2 according to Henry’s Law, and incorporating empirical data
from field measurements of soil moisture content, pH, and temperature changes, showed that an
abiotic mechanism of CO2 uptake and release could explain changes in subsoil CO2 concentrations.
However, the twice-daily sampling regime did not resolve diel changes or rates of change in surface
CO2 fluxes and subsurface CO2 concentration and δ13CCO2 profiles.
During the 2009/10 austral summer, a 48-h time-series of surface CO2 fluxes and subsurface CO2
concentration and δ13CCO2 profiles were measured simultaneously at Sites A and B, at 4-hourly
intervals. The increased sampling resolution revealed dynamic changes in subsurface soil CO2
iii
concentration and δ13CCO2 profiles as soil temperatures varied throughout the diel cycle. At Site B,
highly depleted surface CO2 fluxes were explained by dynamic fractionation effects, and could not be
attributed to biological respiration. Short-lived (4 h) periods of steady-state efflux and influx showed
CO2 production (exsolution) and consumption (dissolution) of relatively enriched (−5.2‰) and
depleted CO2 (−11.4‰), respectively. This is consistent with the kinetic fractionation expected as a
result of preferential exsolution of 13CO2 and dissolution of 12CO2.
At Site A, static chambers were unreliable for determining surface CO2 flux rates. Nonetheless,
variation in subsurface CO2 concentration and δ13CCO2 profiles was consistent with heterotrophic
respiration dominating soil CO2 production. Unexpectedly, the highest rates of CO2 production
occurred during cooler periods of the day, possibly as a result of cooling-induced condensation
providing biota with water sufficient to stimulate their activity.
An abiotic contribution to Dry Valley soil CO2 fluxes must be considered in future studies of Dry Valley
ecosystem activity and C cycling. Measures of surface CO2 flux at single points in time do not
necessarily capture a solely biological signal.
Keywords: McMurdo Dry Valleys, soil respiration, Antarctic soils, organic C, dynamic fractionation,
CO2 fluxes.
Glossary of terms
δ13C
13
δ13CCO2
13
C isotopic composition
C isotopic composition of CO2 (this can be soil CO2, or CO2 sampled from a
headspace chamber)
δ13CΔCO2
13
ΔT/Δt
rate of change in soil temperature
CT
total dissolved CO2
De
effective diffusivity of CO2 in soil
Site A
representative of a “biological hotspot”; has a contemporary source of organic C
Site B
representative of “typical” Dry Valley soils; has no contemporary organic C source
C isotopic composition of the net surface CO2 flux
iv
Acknowledgements
It has been a great privilege to work in the Dry Valleys of Victoria Land: the magic and
unpredictability of Antarctica always makes for challenging, memorable and rewarding experiences.
Many individuals have helped ensure the success of the field work, and others have provided
excellent advice during the laboratory and write-up phases of this work.
Firstly, I thank my supervisor Tim Clough for his critical evaluation and thought-provoking feedback.
It is unfortunate that Tim was not able to visit the field area, and I hope that one day that may be
possible as part of another project. Thanks also to my associate supervisor, Carol Smith, who was
always enthusiastic about the research, and contributed to the successful field season in 2008/09,
not only in terms of research, but also with her sense of humour and camp culinary skills. Peter
Almond is thanked for his invaluable feedback, and for guiding me through the mathematicallydemanding aspects of this study. Peter also contributed to all aspects of the field work in both
2008/09 and 2009/10. His commitment to carrying gear and taking gas samples at unearthly hours
was outstanding, and his efforts also extended to the creation of “polygon scones” and legendary
lamb roasts. My supervisors must collectively be thanked greatly for their encouragement and
suggestions, and for patiently understanding my need to return to Antarctica during the summer of
2011/12 to work on a different research project investigating factors controlling biodiversity in the
Prince Charles Mountains, Mac. Robertson Land.
Jim Bockheim’s knowledge of Taylor Valley and soils of the Transantarctic Mountains, as well as his
enthusiasm and assistance, was greatly appreciated in the 2008/09 field season. André Eger and
Vicky Nall provided outstanding field assistance in the 2009/10 summer – assisting with gas sampling
at all hours, and always ensuring a healthy mix of melodies were transmitted during the “Drive Time”
radio shows. All of the field work and associated logistics was made possible by the excellent
logistical support provided by Antarctica New Zealand. Shulamit Gordon, Paul Woodgate, Simon
Trotter, Rod Strachan, and the Scott Base summer staff of 2008/09 and 2009/10 are particularly
thanked for their role in ensuring all logistical aspects of the project ran smoothly. Thanks also to Rob
McPhail and his mechanics for our safe and enjoyable flights to and from the field sites.
Staff of the Department of Soil and Physical Sciences also made major contributions to the project,
for which I am very grateful. Amanda Clifford and Leanne Hassall ensured the timely arrival of all
necessary equipment, and Neil Smith spent long hours helping with programming and testing the
data loggers. Leanne is especially thanked for her wise counsel and valued friendship. Roger
Cresswell’s help with gas sampling preparations and analyses was invaluable. Roger worked tirelessly
to ensure the mass spec was running well, with assistance from Manjula Premaratne. Lynne Clucas
v
and Joy Jiao patiently analysed and re-analysed many samples. Thanks also to Rob Sherlock and the
late Graeme Buchan for the opportunities afforded to me during my time as a PhD student, as well as
for their support and encouragement, and many helpful discussions about my work. Janet Bertram,
Amal Torky and Jim Moir are also thanked for their support and encouragement. Thank you also to
my friends and fellow students in the Department for their good company – particularly André Eger,
Vicky Nall, Brendon Malcolm, Nimlesh Balaine, Pranoy Pal and Jen Owens.
Thanks also to Barry Thornton and his team at the Macaulay Land Use Research Institute in
Aberdeen, Scotland for making me very welcome in their laboratory, and for their careful analysis of
many gas samples.
Thank you to Megan Balks for her support throughout my PhD. Megan is always a cheery voice down
the end of the phone, and I am very grateful for her helpful suggestions and advice, as well as for
sparking my Antarctic research adventures by giving me the opportunity to visit Wright Valley as a
Masters student nine years ago. Jenny Webster-Brown also gave helpful suggestions and advice. Ed
Gregorich shared his knowledge of gas sampling methods he had used in the Dry Valleys, and John
Hunt gave advice on chamber design.
Operational support for the project was provided by the Lincoln University Research Fund. I am
extremely grateful for financial assistance from Helicopters New Zealand, the William Machin Trust,
Freemasons New Zealand, the Kate Sheppard Memorial Trust and Kathryn Claridge of Lincoln
Physiotherapy.
Thank you to my dear friends Michele Frey, Jodi Antunovich and Charlotte Fletcher for your
wonderful support and encouragement, and for giving me so many reasons to smile.
To my partner, Peter: thank you so much for your patience, support, understanding and love. Finally,
to my Mum and Dad, and my brother David: I cannot thank you enough for your love, advice and
encouragement. You taught me to dream, and have always supported me and encouraged me to
make and take opportunities.
vi
Table of Contents
Abstract ...................................................................................................................................... iii
Acknowledgements ...................................................................................................................... v
Table of Contents ....................................................................................................................... vii
List of Tables ............................................................................................................................... xi
List of Figures ............................................................................................................................ xiii
Chapter 1 Introduction ................................................................................................................. 1
1.1 Background .................................................................................................................................... 1
1.2 Aim and objectives ......................................................................................................................... 2
1.3 Field work....................................................................................................................................... 3
1.4 Thesis structure.............................................................................................................................. 3
1.5 Publications and conference presentations resulting from this thesis ......................................... 4
1.5.1 Peer-reviewed journal articles to date ............................................................................. 4
1.5.2 Selected conference presentations .................................................................................. 4
Chapter 2 Literature review .......................................................................................................... 5
2.1 Introduction ................................................................................................................................... 5
2.2 Soil CO2 dynamics and effects on carbon isotopic composition .................................................... 6
2.2.1 CO2 sampling using headspace chambers......................................................................... 6
2.2.2 Isotopic fractionation ........................................................................................................ 7
2.2.3 Biological respiration ...................................................................................................... 12
2.2.4 Dissolution of CO2 in water ............................................................................................. 13
2.2.5 Carbon isotopic fractionation in the carbonate system ................................................. 15
2.3 McMurdo Dry Valleys .................................................................................................................. 16
2.3.1 Climate ............................................................................................................................ 17
2.3.2 Hydrology ........................................................................................................................ 18
2.3.3 Geomorphology .............................................................................................................. 19
2.3.4 Soils ................................................................................................................................. 20
2.3.5 CO2 fluxes ........................................................................................................................ 27
2.4 Research gaps .............................................................................................................................. 29
Chapter 3 Abiotic processes dominate CO2 fluxes in Antarctic soils .............................................. 31
3.1 Abstract ........................................................................................................................................ 31
3.2 Introduction ................................................................................................................................. 32
3.3 Materials and Methods................................................................................................................ 33
3.3.1 Study area ....................................................................................................................... 33
3.3.2 Soil temperature and moisture content measurements ................................................ 37
3.3.3 Soil CO2 sampling ............................................................................................................ 37
3.3.4 Soil sampling, physical and chemical analyses................................................................ 39
3.3.5 Statistical analyses .......................................................................................................... 40
3.4 Results .......................................................................................................................................... 40
3.4.1 Environmental variables ................................................................................................. 40
3.4.2 Soil physical and chemical properties ............................................................................. 43
3.4.3 CO2 dynamics .................................................................................................................. 45
3.5 Discussion..................................................................................................................................... 49
vii
3.6
3.7
3.5.1 Soil CO2 uptake and release ............................................................................................ 49
Conclusions .................................................................................................................................. 57
Supplementary information......................................................................................................... 58
3.7.1 Model of abiotic CO2 uptake and release ....................................................................... 58
3.7.2 Calculating the isotopic composition of abiotically-produced CO2 (δ13CCO2-abiotic) ......... 59
3.7.3 Results of principal components analysis ....................................................................... 60
Chapter 4 Quantifying diel variations in biotic and abiotic soil CO2 fluxes, Taylor Valley,
Antarctica (I): Background rates in the absence of a contemporary organic carbon source............ 61
4.1 Introduction ................................................................................................................................. 61
4.2 Materials and methods ................................................................................................................ 62
4.2.1 Study area ....................................................................................................................... 62
4.2.2 Environmental variables ................................................................................................. 62
4.2.3 Soil CO2 sampling and analysis ........................................................................................ 62
4.2.4 Soil sampling, physical and chemical analyses................................................................ 63
4.2.5 Statistical analyses and calculations ............................................................................... 64
4.3 Results .......................................................................................................................................... 69
4.3.1 Environmental variables ................................................................................................. 69
4.3.2 CO2 dynamics .................................................................................................................. 71
4.3.3 Linking surface and subsurface CO2 dynamics ................................................................ 76
4.4 Discussion..................................................................................................................................... 82
4.4.1 Diel variability in surface and subsurface storage CO2 flux rates ................................... 82
4.4.2 Soil gas transport ............................................................................................................ 86
4.4.3 Sources of CO2 ................................................................................................................. 89
4.4.4 Data reliability ................................................................................................................. 94
4.4.5 Implications ..................................................................................................................... 94
4.5 Summary and conclusions ........................................................................................................... 96
Chapter 5 Quantifying diel variations in biotic and abiotic soil CO2 fluxes, Taylor Valley,
Antarctica (II): The influence of a contemporary organic carbon source........................................ 99
5.1 Introduction ................................................................................................................................. 99
5.2 Materials and methods .............................................................................................................. 100
5.2.1 Study area ..................................................................................................................... 100
5.2.2 Environmental variables ............................................................................................... 100
5.2.3 Soil CO2 sampling and analysis ...................................................................................... 101
5.2.4 Soil sampling, physical and chemical analyses.............................................................. 101
5.2.5 Statistical analyses and calculations ............................................................................. 102
5.3 Results ........................................................................................................................................ 105
5.3.1 Environmental variables ............................................................................................... 105
5.3.2 CO2 dynamics ................................................................................................................ 107
5.3.3 Linking surface and subsurface CO2 dynamics .............................................................. 113
5.4 Discussion................................................................................................................................... 117
5.4.1 Data reliability ............................................................................................................... 121
5.4.2 Implications ................................................................................................................... 122
5.5 Summary and conclusions ......................................................................................................... 124
5.6 Supplementary information....................................................................................................... 125
5.6.1 Detailed description of surface and subsurface CO2 dynamics on Day 2 ..................... 125
Chapter 6 Synthesis and Summary .............................................................................................129
6.1 Introduction ............................................................................................................................... 129
viii
6.2
6.3
6.4
Key findings ................................................................................................................................ 130
Implications of research findings ............................................................................................... 133
6.3.1 Comparisons of Dry Valley ecosystem activity ............................................................. 134
6.3.2 Carbon turnover times .................................................................................................. 134
6.3.3 Field and laboratory manipulations .............................................................................. 135
Future research .......................................................................................................................... 136
6.4.1 Laboratory studies ........................................................................................................ 136
6.4.2 Numerical modelling studies......................................................................................... 136
6.4.3 Field soil CO2 flux experimental design ......................................................................... 137
References ............................................................................................................................... 139
Appendix A Publications resulting from this thesis......................................................................149
ix
x
List of Tables
Table 2.1
Table 2.2
Table 2.3
Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 4.1
Table 5.1
Carbon isotope fractionation in the CO2−HCO3−CO3−CaCO3 system under equilibrium
conditions. 13εy/x represents the fractionation of compound y relative to compound x.
Soil temperatures (T) shown are similar to those measured in the McMurdo Dry
Valleys. g = gaseous CO2, a = dissolved CO2, b = dissolved HCO3−, c = dissolved CO32−
ions, s = solid calcium carbonate. TK = T (°C) + 273.15 K. .................................................. 15
Changes in soil morphology and surface weathering features with increasing
weathering (Campbell and Claridge, 1975). ...................................................................... 22
Morphogenetic stages of accumulation of soluble salts in Cold Desert soils
(Bockheim, 1990). .............................................................................................................. 23
Surface soil CO2 fluxes and rate of change in δ13CCO2 during surface CO2 flux sampling
at Sites A and B, Taylor Valley. .......................................................................................... 48
Carbon isotope fractionation in the CO2−HCO3−CO3−CaCO3 system under equilibrium
conditions. 13εy/x represents the fractionation of compound y relative to compound x.
Soil temperatures (T) shown are similar to those measured during the sampling
period. g = gaseous CO2, a = dissolved CO2, b = dissolved HCO3−, c = dissolved CO32−
ions, s = solid calcium carbonate. TK = T (°C) + 273.15 K. .................................................. 54
Proportion of the change in CO2 concentration and isotopic composition attributable
to biotic processes (fbiotic) at Site A, 5−15 cm depth. This depth has the greatest shift
in CO2 concentration between cold and warm periods (Figure 3.10A) and is the most
likely place in which biological respiration may occur. The effects of changes in
assumed δ13CCO2-abiotic values on fbiotic are shown. .............................................................. 56
Eigenvector coefficients and component correlations for the first three principal
components. Highlighted cells represent values higher than Mardia’s criterion (0.7
times the highest coefficient for the component) for the eigenvector coefficients.
Highlighted cells in the component correlations columns show variables that are well
described by the components. .......................................................................................... 60
Surface soil CO2 fluxes at Site B, Taylor Valley. ................................................................. 73
Surface soil CO2 fluxes at Site A, Taylor Valley. ............................................................... 108
xi
xii
List of Figures
Figure 1.1
Field camp beside Spaulding Pond, eastern Taylor Valley. Howard Glacier is on the
left hand side of the photograph (taken 27 December 2009)............................................. 3
Figure 2.1 The effect of pH on the distribution of dissolved inorganic carbon species in solution. .. 14
Figure 2.2 The polar desert landscape of the McMurdo Dry Valleys. Clockwise from top: alpine
glaciers and ice-covered Lake Bonney, central Taylor Valley; polygonal patterned
ground, lower Taylor Valley; ephemeral stream, lower Taylor Valley. ............................. 16
Figure 3.1 Location of Sites A and B, Taylor Valley, Antarctica. ......................................................... 33
Figure 3.2 Site A, lower Taylor Valley. Clockwise from top: view west to gas sampling area and
Howard Glacier; headspace chambers with Spaulding Pond in the background; algae
accumulating at the edge of Spaulding Pond. ................................................................... 35
Figure 3.3 Site B, lower Taylor Valley. Top: view west showing gas sampling chambers and
tubes. Bottom: view south-east showing soil pit on broad moraine crest. Site A is
located next to the lake in the distance on the left hand site of the photo...................... 36
Figure 3.4 Soils (Typic Haploturbels) at Site A and Site B. At the time of sampling (late December
2008), soils at Sites A and B were ice-cemented below 35 and 26 cm, respectively.
The profile shown for Site B has been deepened with a concrete breaker and is
moistened from melting of ice-cemented permafrost...................................................... 37
Figure 3.5 Static headspace chamber and subsurface sampling tubes, Site B, Taylor Valley. ........... 38
Figure 3.6 Soil temperature (A) and moisture content (B) during gas sampling periods, Site A
Taylor Valley. Note: data at 5 cm depth were only recorded during one sampling
period due to a faulty probe.............................................................................................. 41
Figure 3.7 Soil temperature (A) and moisture content (B) during gas sampling periods, Site B
Taylor Valley. Note: this site was only sampled during one “cold” period due to an
error in determining the timing of the coldest part of the day. ........................................ 42
Figure 3.8 Depth profiles of A) soil inorganic C, B) soil organic C, and C) C isotopic composition
of soil organic C at Sites A and B, Taylor Valley. Error bars are ±1 standard deviation; n
= 2. ..................................................................................................................................... 44
Figure 3.9 Depth profiles of A) soil pH, and B) electrical conductivity at Sites A and B, Taylor
Valley. Error bars are ±1 standard deviation; n = 2. .......................................................... 45
Figure 3.10 Soil profile CO2 concentration (A) and δ13CCO2 (B) at Sites A and B, Taylor Valley. Data
shown represent averages for warm and cold sampling periods (n = 6 for 0 cm
samples and n = 12 at all other depths, except for the cold sampling period at Site B,
where n = 3 for 0 cm samples and n = 6 at all other depths). Error bars represent the
SEM. ................................................................................................................................... 46
Figure 3.11 Ratios of simulated [CO2]warm: [CO2]cold as influenced by soil moisture content and soil
solution pH. Ratios are calculated using equation 3.3. In (A) Twarm = 286 K and Tcold =
274 K which are similar to soil temperatures measured at 5 cm depth, and in (B) Twarm
= 278.3 K and Tcold = 277.3 K, similar to soil temperatures measured at warm and cold
periods, respectively, at 15 cm depth in the study area. When the ratio of [CO2]warm:
[CO2]cold is 1, there is zero potential for CO2 to be absorbed. This occurs when soil
moisture is so low that no CO2 can be dissolved. Higher ratios represent greater
changes in soil CO2 concentrations between warm and cold periods, and thus greater
uptake and release of CO2 by soils. ................................................................................... 52
Figure 3.12 Comparison between measured and simulated [CO2]warm: [CO2]cold ratios between
warm and cold sampling periods for Sites A and B, Taylor Valley..................................... 53
Figure 4.1 Morphological comparison between soil pits adjacent to gas sampling sites at Site B
in the 2008/09 and 2009/10 austral summers. At the time of sampling, the soils
(Typic Haploturbels) were ice-cemented below 26 and 30 cm depth in 2008/09 and
2009/10, respectively. Both profiles have been deepened with a concrete breaker
and are moistened from melting of ice-cemented permafrost. ....................................... 64
xiii
Figure 4.2
Soil temperature (A) and volumetric moisture content (B) during the 48-hour period
over which gas samples were taken, Site B, Taylor Valley. Data plotted for moisture
content are 30 min. moving averages. The narrow vertical grey boxes denote the
periods during which gas samples were taken.................................................................. 70
Figure 4.3 (A) Average significant surface and subsurface soil CO2 flux rates at Site B, Taylor
Valley. Error bars represent the standard error of the mean. Subsurface soil CO2
storage fluxes are shown as grey bars which represent the net CO2 flux over the 4-h
period between sampling. (B) Isotopic composition (δ13CΔCO2) of measured and
predicted surface CO2 fluxes at Site B, Taylor Valley. Predicted values are calculated
based on subsurface CO2 concentration and δ13CCO2 gradients at the time of sampling,
according to equation 4.26. Error bars represent the standard error of the mean. The
dashed horizontal line represents the average δ13CCO2 of soil CO2 (−8.8‰). (C) Soil
temperature at 5 and 15 cm depth (left hand axis) during the gas sampling period,
Site B, Taylor Valley. Wind speed data (right hand axis) are from Site A, Taylor Valley.
(D) Rate of change in soil temperature at 5 and 15 cm depth during the gas sampling
period, Site B, Taylor Valley. Data plotted are 30 min moving averages. The dashed
vertical lines correspond to the four-hourly intervals at which gas sampling began. ...... 72
Figure 4.4 Soil profile CO2 concentrations and δ13CCO2 values at Site B, Taylor Valley. (A) and (B)
represent data from Day 1 and Day 2, respectively. Data points are averages from
each sampling time (n = 1 or 2 for 0 cm samples. At all other depths, n = 4). Error bars
represent the standard error of the mean. ....................................................................... 75
Figure 4.5 Relationship between average temperature over the surface flux sampling period and
surface CO2 flux rates at Site B, Taylor Valley.................................................................... 83
Figure 4.6 Relationship between the average rate of change in soil temperature over the
surface flux sampling period at 5 cm and 15 cm depth with surface CO2 flux rates at
Site B, Taylor Valley. .......................................................................................................... 83
Figure 4.7 Net cumulative surface CO2 fluxes at Site B, Taylor Valley, over consecutive 24-h
periods beginning at 14:30 h on Day 1. Maximum and minimum net fluxes represent
uncertainties calculated from the standard error of the mean. ....................................... 84
Figure 4.8 Relationship between the rate of change in soil temperature at 5 cm and 15 cm
depth with subsurface storage CO2 flux rates at Site B, Taylor Valley. The solid black
lines represent linear relationships with equations as follows. At 5 cm depth, y =
0.0059x + 0.0002. At 15 cm depth, y = 0.0047x - 0.0001. ................................................. 85
Figure 4.9 Relationship between subsurface CO2 concentration gradient and surface CO2 flux on
combined data from Days 1 and 2, Site B, Taylor Valley. The solid black line
represents a linear relationship: y = 0.032x + 8E-07; the slope of the line represents
the effective diffusivity of CO2 in the soil. Anomalous data from 06:30 h on Day 2
were excluded. .................................................................................................................. 86
Figure 4.10 Relationship between wind speed and surface CO2 flux rates at Site B, Taylor Valley.
The solid black line represents a linear relationship: y = 0.0189x − 0.0353. ..................... 87
Figure 4.11 Schematics showing the influence of CO2 concentration and δ13CCO2 gradients on the
δ13CΔCO2 of surface CO2 fluxes and on the δ13CCO2 of soil CO2. The letters A, B, C, and D
refer to various scenarios described in the text. ............................................................... 90
Figure 4.12 Relationship between the δ13CΔCO2 of measured and modelled surface CO2 fluxes from
Days 1 and 2, Site B, Taylor Valley. The anomalous data point from 06:30 h on Day 2
was excluded from this plot and from the linear regression. The solid black line is a
linear relationship relating x (measured) and y (modelled) values as follows: y = 0.94x
– 1.42. ................................................................................................................................ 92
Figure 5.1 Morphological comparison between soil pits adjacent to gas sampling sites at Site A
in the 2008/09 and 2009/10 austral summers. At the time of sampling, both soils
(Typic Haploturbels) were ice-cemented below 35 cm depth. The moistening of the
upper ~ 2 cm of the 2009/10 profile is from melting of recent snowfall. ....................... 102
Figure 5.2 Soil temperature (A) and volumetric moisture content (B) during the 48-h period over
which gas samples were taken, Site A, Taylor Valley. Data plotted for moisture
xiv
Figure 5.3
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
content are 30 min. moving averages. The narrow vertical grey boxes denote the
periods during which gas samples were taken................................................................ 106
Selection of CO2 time-series data from surface headspace chambers, Site A, Taylor
Valley, showing variation in CO2 concentration over the 45-min sampling period. Note
that data from each sampling period are plotted on different CO2 concentration
scales................................................................................................................................ 109
(A) Average significant measured surface CO2 flux rates, predicted surface CO2 flux
rates, and subsurface soil CO2 storage flux rates at Site A, Taylor Valley. Error bars
represent the standard error of the mean. Subsurface soil CO2 storage fluxes are
shown as grey bars, which represent the net CO2 flux over the 4-h period between
sampling. (B) Isotopic composition (δ13CΔCO2) of measured and predicted surface CO2
fluxes at Site A, Taylor Valley. Predicted values are calculated based on subsurface
CO2 concentration and δ13CCO2 gradients at the time of sampling. Error bars represent
the standard error of the mean. The dashed horizontal line represents the average
δ13CCO2 of ambient atmospheric CO2 (−8.8‰). (C) Soil temperature at 5 and 15 cm
depth (left hand axis) and wind speed (right hand axis) during the gas sampling
period, Site A, Taylor Valley. (D) Rate of change in soil temperature at 5 and 15 cm
depth during the gas sampling period, Site A, Taylor Valley. Data plotted are 30 min
moving averages. The dashed vertical lines correspond to the four-hourly intervals at
which gas sampling began. .............................................................................................. 110
Soil profile CO2 concentration and δ13CCO2 at Site A, Taylor Valley. (A) and (B)
represent data from Day 1 and Day 2, respectively. Data shown represent averages
from each sampling time (n = 1 or 2 for 0 cm samples. At all other depths, n = 4).
Error bars represent the standard error of the mean. .................................................... 112
Relationship between average temperature over the surface flux sampling period and
predicted surface CO2 flux rates at Site A, Taylor Valley. Regressions are not
significant (p > 0.1). ......................................................................................................... 118
Relationship between the average rate of change in soil temperature over the
surface flux sampling period at 5 cm and 15 cm depth with predicted surface CO2 flux
rates at Site A, Taylor Valley. Regressions are not significant (p > 0.1)........................... 118
xv
Chapter 1
Introduction
1.1 Background
The McMurdo Dry Valleys are the largest of the ice-free areas in Antarctica, and comprise lowcomplexity polar desert ecosystems within one of the most inhospitable environments on Earth. Dry
Valley terrestrial ecosystems exist under conditions considered to be biologically extreme: liquid
water is scarce, winds can be strong, there are large diel and seasonal variations in temperature, and
extremes in solar radiation which range from 0-h daylight during the dark winter months, to 24-h
daylight during the austral summer. Soils are generally saline, and contain low levels of carbon (C)
and nitrogen (N). However, despite these largely unfavourable conditions for life, biological
communities are widely distributed throughout the Dry Valleys.
The harsh conditions under which Dry Valley biological communities exist dictate that ecosystems
are relatively simple, thus providing a unique opportunity to test theories of trophic interactions and
factors controlling community structure and function (Wall and Virginia, 1999). The advent of
modern molecular genetic sequencing techniques has also enabled questions about how
environmental factors influence biodiversity and ecosystem functioning to be addressed. As small
changes in environmental conditions can have profound effects on Dry Valley ecosystems (Fountain
et al., 1999), such studies are often predicated on providing a better understanding of the
distribution, diversity and activity of biota in order to facilitate predictions about how community
structure and function will be affected by future environmental change. One such measure of
biological activity is soil respiration.
Previous studies of soil respiration in the Dry Valleys have provided insight into the physical controls
on soil carbon dioxide (CO2) flux rates (Parsons et al., 2004), and the effects of simulated climate
change on C cycling (Ball et al., 2009). In addition, soil respiration rates have been used to calculate
the turnover time of the soil organic C pool (Burkins et al., 2001; Elberling et al., 2006), and to
provide a basis for comparing ecosystem function between sites (Barrett et al., 2006b). Long-term
records of soil respiration rates have been proposed as potentially providing one of the most
sensitive indicators of ecosystem response to environmental change (Barrett et al., 2006c).
Recent studies of Dry Valley soil CO2 fluxes have identified that surface CO2 flux rates follow diel
variations, with positive fluxes measured during warmer parts of the day, and negative fluxes
measured during cooler periods (Parsons et al., 2004; Ball et al., 2009). The negative CO2 fluxes have
1
no biological explanation, and, as changes in soil temperature have been shown to account for
almost half of the variation in CO2 flux rate (Ball et al., 2009), it has been suggested that Henry’s Lawcontrolled dissolution of CO2 in soil water may account for the diel variability observed (Parsons et
al., 2004; Ball et al., 2009). The potential abiotic contribution to soil CO2 fluxes (via exsolution of CO2
as soil temperatures increase) means that surface CO2 flux rates cannot be assumed to solely
represent biological CO2 production.
An abiotic contribution to soil CO2 fluxes has potentially profound implications for the interpretation
of Dry Valley CO2 fluxes, particularly in terms of quantifying biological activity and rates of C cycling. A
significant abiotic contribution to soil CO2 fluxes would mean that previous measurements of soil
respiration rate are overestimates of the rate of biological CO2 production, and hence turnover times
calculated from such fluxes would be underestimates. Furthermore, as any abiotic component is
likely to vary according to local site conditions, using surface CO2 flux rates as a means of comparing
ecosystem activity between sites may not provide an accurate representation of differences in biotic
CO2 production. Single measures of soil CO2 flux rate may comprise both biotic and abiotic
components, thus to capture an absolute biological signal would require repeated measurements
over sufficiently long timescales so as to integrate any cyclical abiotic effects. Further research is
needed to partition the biotic and abiotic components of soil CO2 fluxes, from which biotic and
abiotic rates of CO2 production can be quantified.
1.2 Aim and objectives
This study aimed to characterise and quantify the detailed variations in surface CO2 flux rates and
subsurface CO2 concentration and 13C isotopic composition (δ13CCO2) profiles at two sites with
contrasting organic C inputs. This research builds on previous work on Dry Valley soil CO2 fluxes by
using high-resolution measurements of surface CO2 flux rate and, for the first time, subsurface CO2
concentration and δ13CCO2 profiles in order to link surface and subsurface soil CO2 dynamics, and to
discriminate between biotic and abiotic sources of CO2.
The study comprised both field and laboratory work in order to achieve the following objectives:
•
To test whether abiotic mechanisms of CO2 production and consumption can explain
observed diel variability in surface CO2 flux rates (Chapter 3);
•
To determine the relative importance of abiotic factors on soil CO2 production and
consumption at a site with low levels of organic C, which is representative of “typical” Dry
Valley soils (Chapter 4);
2
•
To partition the biotic and abiotic components of surface CO2 fluxes at a site with relatively
high levels of organic C, characteristic of a biological “hotspot” (Chapter 5).
1.3 Field work
The results presented in this thesis are based on field work carried out during the 2008/09 and
2009/10 austral summers. The main campsite from which this work was undertaken was located ~
800 m east of Howard Glacier, in eastern Taylor Valley (Figure 1.1).
Figure 1.1
Field camp beside Spaulding Pond, eastern Taylor Valley. Howard Glacier is on the left
hand side of the photograph (taken 27 December 2009).
1.4 Thesis structure
The research is presented in three main results chapters which follow a literature review (Chapter 2)
that provides the background to and rationale for the study. The first results chapter (Chapter 3) is
written as a manuscript, and has been published in Soil Biology and Biochemistry. Correspondence
prompted by this paper is to be published as a Letter to the Editor (Elberling et al., 2013), along with
the subsequent response, in Soil Biology and Biochemistry. The published paper, together with the
comment and response, are included as Appendix A. Following some alterations to the style in which
results are described, the second and third results chapters (Chapters 4 and 5) will also be submitted
to Soil Biology and Biochemistry. A synthesis and summary of findings is presented in Chapter 6.
3
1.5 Publications and conference presentations resulting from this thesis
1.5.1 Peer-reviewed journal articles to date
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., 2012. Abiotic processes dominate CO2 fluxes
in Antarctic soils. Soil Biology & Biochemistry, 53: 99−111. doi:10.1016/j.soilbio.2012.04.027.
This article resulted in a Letter to the Editor (Elberling et al., 2013; Appendix A) and a reply as follows:
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., 2013. Reply to Elberling et al.’s (2013)
comments on “Abiotic processes dominate CO2 fluxes in Antarctic soils” (Soil Biol. Biochem. 53,
99−111). Soil Biology & Biochemistry, In Press, http://dx.doi.org/10.1016/j.soilbio.2013.10.039.
1.5.2 Selected conference presentations
Shanhun, F., Almond, P., Clough, T., Smith, C., 2013. Diel variation in the biotic and abiotic
components of soil CO2 fluxes. XIth SCAR Biology Symposium, Barcelona, Spain. 15−19 July 2013.
Shanhun, F., Almond, P., Clough, T., Smith, C., 2011. Abiotic processes dominate soil CO2 fluxes in
Taylor Valley. 2011 Annual Antarctic Conference, University of Waikato, Hamilton. 4−6 July 2011.
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., Bockheim, J.G., 2009. Isotopic composition of
soil CO2 in Antarctica: Implications for a pedogenic carbonate paleotemperature proxy. 2009 Lincoln
University Postgraduate Conference. 31 August – 1 September 2009.
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., Bockheim, J.G., 2009. Reconstructing past
temperatures in ice-free areas: Development of a pedogenic carbonate proxy. 2009 Annual Antarctic
Conference, University of Auckland. 1−3 July 2009.
4
Chapter 2
Literature review
2.1 Introduction
The McMurdo Dry Valleys are the largest contiguous ice-free area in Antarctica, and are home to
organisms that can withstand conditions that are considered to be biologically extreme: strong
winds, low moisture availability, low nutrient availability, high salinity, large daily and seasonal
variations in temperature, and extremes in solar radiation that range from total darkness during the
winter months, to 24-h daylight during the austral summer. The low-complexity ecosystems found
within the McMurdo Dry Valleys provide a unique natural laboratory for answering fundamental
questions about the survival, distribution, and activity of biological communities in the absence of
vascular plants. These ecosystems provide an opportunity to study abiotic controls on ecosystem
structure and function, as well as their sensitivity to changes in environmental conditions (Fountain
et al., 1999; Wall and Virginia, 1999). Consequently, the McMurdo Dry Valleys are an important
indicator environment from which predictions of future change can be made.
Recent research has focused on understanding the factors controlling the distribution and diversity
of biota throughout the Dry Valleys and beyond (Lee et al., 2012; Magalhães et al., 2012; Stomeo et
al., 2012; Van Horn et al., 2013), and previous studies have aimed to quantify rates of soil respiration,
with such rates being used to inform calculations of soil organic carbon (C) turnover time (Burkins et
al., 2001; Barrett et al., 2006a; Elberling et al., 2006), and to compare ecosystem function between
sites (Barrett et al., 2006b). These previous studies were predicated on the assumption that, as in
temperate environments, surface carbon dioxide (CO2) fluxes are representative of biological CO2
production. However, as abiotic factors largely control biological distribution, diversity and activity in
the Dry Valleys (Barrett et al., 2004; Hogg et al., 2006; Lee et al., 2012; Magalhães et al., 2012;
Stomeo et al., 2012; Van Horn et al., 2013), this assumption may not necessarily apply to dry valley
ecosystems, and needs to be tested. Furthermore, recent studies investigating Dry Valley soil CO2
fluxes at regular intervals over several diel cycles have identified negative CO2 flux rates (Parsons et
al., 2004; Ball et al., 2009), with the rate of change in soil temperature, rather than variation in
absolute temperature, accounting for almost half of the variability in surface CO2 flux rates (Ball et
al., 2009). Since negative CO2 flux rates have no biological explanation, it has been suggested that
Henry’s Law-controlled dissolution and exsolution of CO2 may account for the diel variability in
surface CO2 flux rates (Parsons et al., 2004; Ball et al., 2009). This hypothesis is, as yet, untested, and
forms the basis for the research presented in this study.
5
This review examines literature pertinent to measurement and interpretation of soil CO2 fluxes, and
provides a basis for understanding CO2 dynamics in Dry Valley ecosystems.
2.2 Soil CO2 dynamics and effects on carbon isotopic composition
This thesis examines the rate and C isotopic composition (δ13CCO2) of soil CO2 fluxes and depth
profiles of CO2 concentration and δ13CCO2 in a polar desert environment. A short section on soil gas
sampling and factors affecting the isotopic composition of soil CO2 is presented here to provide the
necessary background to understand the results presented in Chapters 3, 4, and 5. More detailed
discussion of the current knowledge of soil CO2 fluxes in the McMurdo Dry Valleys follows in section
2.3.5.
2.2.1 CO2 sampling using headspace chambers
Surface CO2 flux rates are typically determined via the use of headspace chambers. Such chambers
can be manually sampled, where gas samples are extracted from the chamber and injected into a vial
for later analysis, or coupled to automated flux measurement instruments. As static chambers were
employed in this study, this review is limited to their design and potential limitations. Further
discussion of additional chamber types and their limitations, as derived from a comprehensive
modelling study, can be found in Creelman et al. (2013).
Static chamber design and limitations
Static chambers, which accumulate CO2 in a sealed headspace above the soil surface, enable
calculation of surface CO2 flux rates based on a time-series of changes in CO2 concentration. Such
chambers are closed, thus representing a non-steady-state measurement technique, as they change
the natural dynamics of CO2 diffusion into the free atmosphere. Consequently, any increase (or
decrease) in CO2 concentration within the headspace chamber creates a feedback effect, which
distorts the subsurface soil CO2 concentration gradient (Healy et al., 1996; Conen and Smith, 2000).
This leads to increased storage of gas within the profile (Conen and Smith, 2000), and lateral
diffusion of CO2 (Creelman et al., 2013), both of which cause underestimation of the true surface CO2
flux rate. The underestimation of surface CO2 flux rates calculated using static chamber methods can
be up to 35% (Rayment, 2000; Pumpanen et al., 2004), although individual site factors and sampling
design have a large influence on the likely errors in measurement.
The feedback effect of static chambers on subsurface CO2 concentration gradients is most
pronounced in dry, porous soils with high gas diffusivity (Davidson et al., 2002; Nickerson and Risk,
2009b; Creelman et al., 2013), although chamber height and CO2 production rate (Healy et al., 1996;
Conen and Smith, 2000), depth of insertion (Davidson et al., 2002; Creelman et al., 2013), and
6
duration of deployment (Davidson et al., 2002; Creelman et al., 2013) are all factors that can
moderate the degree of distortion of the subsurface concentration gradient and associated lateral
diffusion. As the soils in the study area are characteristically dry and porous, they are highly
susceptible to lateral diffusion and associated underestimation of surface CO2 flux rates if CO2
concentrations in surface chambers increase substantially. Even if CO2 production rates in dry,
porous soils are low and do not cause substantial accumulation of CO2 in headspace chambers, flux
rates may still be underestimated owing to the relatively large volume of gas that can be stored in
the soil beneath the chamber (Rayment, 2000).
2.2.2 Isotopic fractionation
The stable isotopes of C, 12C and 13C, constitute 98.89% and 1.11% of the C atoms on Earth,
respectively (Fry, 2006). The partitioning (or “fractionation”) of the two stable isotopes of C, on the
basis of differences in their mass in response to physical or biochemical processes, enables them to
be used as a natural tracer. For example, isotopic fractionation in biological systems, such as that
which occurs during photosynthesis, enables the 13C isotopic composition (δ13C) of plant tissue or of
respired CO2 to be used to determine the photosynthetic pathway used in assimilating C. Different
types of fractionation may occur, and these are discussed below in the context of soil CO2 dynamics.
Notation and nomenclature
As absolute isotopic ratios are often small and difficult to measure directly, the isotopic ratios (δ) of
measured samples are reported in units of per mil (‰), relative to an international standard:
𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝛿𝛿𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = �
� × 1000
𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
(2.1)
where Rsample and Rstandard are the ratios of heavy to light isotopes in the sample and standard,
respectively.
Consistent with other literature (e.g. Cerling et al., 1991), ‘soil-respired CO2’ represents the flux of
CO2 passing through a soil, whereas ‘soil CO2’ refers to the CO2 at any given depth within the soil.
Soil-respired CO2 has an isotopic composition (δ13CCO2) that reflects the net production of CO2, but the
δ13CCO2 of soil CO2 reflects a mixture between atmospheric CO2 and any CO2 produced within the soil,
and is not constant with depth (Cerling, 1984). In most systems, soil-respired CO2 is of biological
origin, arising from microbial metabolism of C compounds and root respiration (Cerling, 1984), but in
environments where soil-respired CO2 fluxes are low, net surface fluxes may also include a significant
component of CO2 produced as a result of abiotic processes (e.g. Parsons et al., 2004; Ball et al.,
2009).
7
Equilibrium fractionation
Equilibrium fractionation is the mass-dependent partitioning of isotopes among various species or
compounds in a system at chemical (and thermodynamic) equilibrium. As such, the forward and
backward reaction rates of any particular isotope are identical (Kendall and Caldwell, 1998), and the
distribution of isotopes between compounds is temperature dependent (Craig, 1953). The
temperature-dependent fractionation of isotopes provides a means by which records of past
temperatures can be constructed, as the degree of fractionation can be used to determine the
temperature at which a persistent compound formed.
Kinetic fractionation
Kinetic fractionation is the mass-dependent partitioning of isotopes resulting from incomplete or
unidirectional processes (Mook, 2000; Hoefs, 2009). Kinetic fractionation is associated with the
diffusive transport of gases, and arises from the different diffusivities of different isotopes. During
diffusion of CO2, the isotopically lighter 12CO2 molecule moves faster than the heavier 13CO2 molecule,
with the difference in the diffusivities of the 12CO2 and 13CO2 isotopologues causing a transportrelated fractionation. The diffusion coefficients of 12CO2 and 13CO2 differ by 4.4‰ (Craig, 1953), as
given by:
1
𝐷𝐷( 12𝐶𝐶𝑂𝑂2 )
𝑀𝑀( 12𝐶𝐶𝐶𝐶2 ) + 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 𝑀𝑀( 13𝐶𝐶𝐶𝐶2 ) 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 2
=
�
� = 1.0044
𝐷𝐷( 13𝐶𝐶𝐶𝐶2 )
𝑀𝑀( 12𝐶𝐶𝐶𝐶2 ) 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 𝑀𝑀( 13𝐶𝐶𝐶𝐶2 ) + 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎
(2.2)
where D(12CO2) and D(13CO2) are the diffusion coefficients for 12CO2 and 13CO2, respectively, and
M(12CO2), M(13CO2) and Mair are the atomic masses of 12CO2, 13CO2 and average air, respectively. This
relationship dictates that, in diffusion-driven natural systems, the δ13CCO2 of soil CO2 must be at least
4.4‰ enriched relative to soil-respired CO2 (Cerling et al., 1991).
In a steady-state system, the transport-related fractionation during diffusion is referred to as kinetic
fractionation. For example, in a soil environment with a constant soil respiration rate at a given
δ13CCO2 value, and a constant atmospheric boundary value, the δ13CCO2 of soil CO2 will be at least
4.4‰ enriched with respect to the δ13CCO2 of soil-respired CO2 (Cerling et al., 1991; c.f. error in
Midwood and Millard, 2011). This is because the lighter 12CO2 isotopes move through the soil faster
than the heavier 13CO2 isotopes, resulting in a surface flux that is more depleted than the soil CO2. In
a system at steady-state, the CO2 concentration and δ13CCO2 profiles remain constant (although the
absolute values will vary with depth). Based on gradients of the individual 12CO2 and 13CO2
isotopologues, these profiles dictate that CO2 is emitted from the soil surface at the rate of soil
respiration and with the same δ13CCO2 as that of the soil-respired CO2.
8
The degree of isotopic enrichment of soil CO2 also depends on the soil respiration rate (Cerling, 1984;
Cerling et al., 1991). When the soil respiration rate and soil CO2 concentration is high, the enrichment
of soil CO2 relative to soil-respired CO2 almost exactly matches the theoretical enrichment of 4.4‰
(Cerling et al., 1991). At low soil respiration rates, the difference between the δ13CCO2 of soil CO2 and
soil-respired CO2 is > 4.4‰, due to the enhanced diffusion of 13CO2 from the isotopically heavy
atmosphere into the soil. Low soil respiration rates mean that the soil atmosphere is less depleted
(than in a soil with a high respiration rate) due to a smaller input of isotopically light CO2 produced by
respiration.
Dynamic fractionation
Dynamic fractionation, a term coined by Nickerson and Risk (2009c), describes isotopic fractionation
that occurs in non-steady-state environments. It is similar to kinetic fractionation in that it is also
transport-related, but it results from transient changes in environmental variables driving changes in
12
CO2 and 13CO2 gradients (Nickerson and Risk, 2009c), rather than just differences in the diffusivities
of 12CO2 and 13CO2.
In non-steady-state (dynamic) systems, the theoretical kinetic fractionation of 4.4‰ is unlikely to be
observed (Risk and Kellman, 2008) because the dynamic nature of the environment dictates that
there is a non-infinite source or sink of gas moving towards, but never reaching, equilibrium. Instead,
transient changes in environmental variables such as soil temperature and moisture content drive
changes in biotic and abiotic CO2 production rates, which alter gradients of 12CO2 and 13CO2. Such
changes in the 12CO2 and 13CO2 gradients, which are amplified by the rate of change in the 13CO2
gradient always being slower due to the low natural abundance of 13CO2, lead to transient changes in
the isotopic composition of the soil CO2 flux (Nickerson and Risk, 2009c). For example, an increase in
biological production of isotopically light CO2 acts to steepen the 12CO2 gradient between the soil and
the atmosphere, thus driving 12CO2 out of the profile at a proportionately greater rate than the
slower moving 13CO2, and thereby acting to produce a transiently light surface CO2 flux. Similarly, a
decrease in the rate of biological production of CO2 will result in a decrease in the 12CO2 gradient
between the soil and the atmosphere; this results in proportionately more 13CO2 diffusing out of the
soil, thus producing a transiently heavy surface CO2 flux.
Confounding effects: variation in CO2 production rate
Diurnal variation in the CO2 production rate also influences the isotopic composition of the soil CO2
flux, as shown by Nickerson and Risk (2009c) who used a model to simulate the production and
transport of 12CO2 and 13CO2 in a soil profile. As well as identifying the dynamic fractionation that
occurs (discussed above), their model simulations showed that measured surface CO2 flux rates are
9
out of phase with variations in CO2 production rates, owing to a lagged response which is
predominantly controlled by soil diffusivity.
Furthermore, variation in the isotopic composition of the surface CO2 flux is also out of phase with
variation in CO2 production, such that measured surface flux δ13CCO2 values are not representative of
instantaneous CO2 production, but rather CO2 production that occurred some time before the
measurement was made (Nickerson and Risk, 2009c). Soil diffusivity was again the main property
controlling the time taken for δ13CCO2 values to re-equilibrate following a change in the rate of CO2
production, with higher diffusivities facilitating a more rapid return to equilibrium conditions. High
soil diffusivity values also acted to amplify the degree of deviation from the equilibrium δ13CCO2 value.
Nickerson and Risk’s (2009c) work clearly demonstrates that dynamic fractionation is continuously
present in non-steady-state systems, and requires that, for natural abundance isotope techniques to
be of significant value in discriminating between different sources of CO2, the two sources must
differ isotopically by at least 1‰.
In testing both 1-D and 3-D diffusion scenarios using a model of isotopologue transport as well as
experimental data from static chambers, Nickerson and Risk (2009b) found that the use of Keeling
plots to determine the δ13CCO2 of CO2 being produced is inappropriate in diffusive environments. This
is because differences in the rates of transport of 12CO2 and 13CO2 cause non-linearities in Keeling
plots, which, when relying on linear regression in order to determine the source δ13CCO2, is
unacceptable. Furthermore, in soils with high diffusivity, the effect of non-linearities in Keeling plots
is amplified (Nickerson and Risk, 2009b).
Influence of chamber methodology on δ13CCO2 of soil-respired CO2
Further modelling by Nickerson and Risk (2009a) identified that chamber methodology also
introduces errors in determining the δ13CCO2 of soil-respired CO2. This applies to both steady-state
and non-steady-state chambers, with the magnitude of enrichment in the δ13CCO2 of soil-respired CO2
(relative to the true δ13CCO2 value of soil-respired CO2) ranging from 1 to 4‰. In steady-state
chambers, such as those that use a CO2 sorbent in order to maintain headspace concentration at
ambient atmospheric levels, lateral diffusion of CO2 causes equilibration of the chamber CO2 with
that of soil CO2 at the base of the collar (the lowermost edge of the chamber within the soil).
However, factors such as diffusivity, collar insertion depth, and biological CO2 production rate all
influence the magnitude of the enrichment in measured δ13CCO2 values. In non-steady-state
chambers, such as static chambers in which CO2 accumulates over time (and samples are taken at
regular intervals in order to determine the δ13CCO2 of soil-respired CO2 using a Keeling plot approach),
or purged chambers in which the headspace is initially purged, and then either a single sample is
taken, which is considered to represent the δ13CCO2 of soil-respired CO2, or multiple samples are taken
10
in order to ascertain δ13CCO2 using a Keeling plot approach, both chamber-to-soil feedbacks and
lateral diffusion of CO2 introduce errors in δ13CCO2.
The fact that chamber-to-soil diffusive feedbacks affect δ13CCO2 values is not surprising: increases in
chamber concentration over time act to change the gradients of 12CO2 and 13CO2 between the soil
and the chamber, thus acting to slow overall diffusion rates from the soil to the chamber, as well as
exacerbating the differential diffusion rates of 12CO2 and 13CO2. Lateral diffusion effects also influence
δ13CCO2, with the greatest effects seen when collar lengths are short, and soil diffusivities are high
(Nickerson and Risk, 2009b). As with the steady-state chambers, once lateral diffusion becomes
significant, CO2 in non-steady-state chambers begins to equilibrate with soil CO2 at the depth to
which the collar is inserted (Nickerson and Risk, 2009a). These effects highlight the importance of
utilising a model of 12CO2 and 13CO2 transport in order to accurately interpret isotopic data.
Summary
In dynamic systems, transient changes in environmental variables drive changes in soil CO2
production and δ13CCO2. Consequently, surface CO2 flux and δ13CCO2 measurements alone are
insufficient to determine the timing and magnitude of maxima and minima in soil CO2 production. In
such systems, surface CO2 flux rates are unlikely to be accurate representations of CO2 production
rates due to lags between changes in production and measured surface fluxes. Furthermore,
interpretation of the isotopic composition of the measured surface flux is complicated by lags as well
as dynamic fractionation effects, with the measured δ13CCO2 values not representing the δ13CCO2 of
CO2 being produced in the soil at the time of surface flux measurement.
Interpretation of surface fluxes and their isotopic composition can be aided by installation of soil
profile CO2 sampling tubes. Sampling of subsurface soil CO2 concentration and δ13CCO2 immediately
before or after surface flux sampling provides an instantaneous measure of the gradients of 12CO2
and 13CO2; these gradients can be used to predict surface fluxes and their δ13CCO2. Additionally,
subsurface soil CO2 concentration and δ13CCO2 profiles can be used to determine whether or not there
are periods of steady-state production of CO2, in which case surface flux rates and their isotopic
composition would be representative of the processes happening in the soil over those periods.
Approach used
Despite the problems with the use of static chambers and fractionation effects in non-steady state
environments discussed above, measurements of surface CO2 flux rate and δ13CCO2 were made using
static chambers. This was deemed to be acceptable in this study because surface CO2 flux rates in the
Dry Valleys of Antarctica are known to be low, hence changes in 12CO2 and 13CO2 gradients due to
accumulation of CO2 in headspace chambers are likely to be minimal. Issues of interpretation of
surface fluxes and δ13CCO2 were countered by the installation of subsurface sampling tubes, from
11
which instantaneous profiles of soil CO2 concentration and δ13CCO2 could be used to assess periods of
steady-state and predict surface flux rates and their isotopic composition, based on a simple model
utilising the gradients in 12CO2 and 13CO2. Further details of sampling methods and modelling follow
in the methods sections of Chapters 3 and 4.
2.2.3 Biological respiration
Biological respiration refers to the CO2 released from soils as a result of CO2 production by organisms
and roots (Lloyd and Taylor, 1994) while metabolising C compounds. In the absence of moisture
limitations, the rate of soil respiration increases with increasing soil temperature (Lloyd and Taylor,
1994; Fang and Moncrieff, 2001). The sensitivity of this relationship is amplified at low soil
temperatures (Lloyd and Taylor, 1994; Kirschbaum, 1995), although soil respiration under low
temperatures may be controlled by other mechanisms that are not yet clear (Fang and Moncrieff,
2001). Nonetheless, several confounding effects influence the relationship between soil temperature
and soil respiration. These effects include biotic and abiotic conditions which differ throughout the
soil profile, such as variations in: 1) soil moisture content and CO2 diffusivity; 2) the vertical
distribution of roots and microbes and their individual relationships between respiration rate and
temperature; and 3) the quality and availability of soil organic matter (Subke and Bahn, 2010).
Additionally, differences in the temperature dependence of soil organic C decomposition under
different soil temperature regimes are likely influenced by differences in soil microbial community
composition (Zhu and Cheng, 2011). Therefore, the apparent temperature influence on soil
respiration should be treated cautiously, particularly in cooler environments. A greater
understanding of the interactions between biotic and abiotic processes in different ecosystems is
required in order to make more realistic predictions of soil CO2 efflux under different climate
scenarios (Subke and Bahn, 2010).
Carbon isotopic fractionation in the biological system
Isotopic fractionation during photosynthesis enables differentiation between various biotic
influences on soil CO2 fluxes. The major fractionation of C isotopes associated with photosynthetic
fixation of atmospheric CO2 is a two-step process (Park and Epstein, 1960; O'Leary, 1981). The first
step involves preferential uptake of 12CO2 from the atmosphere (Park and Epstein, 1960), which has a
kinetic fractionation of 4.4‰ (O'Leary, 1981), owing to the differences in diffusivities between the
12
CO2 and 13CO2 isotopologues as they pass through stomata. Subsequent fractionation that occurs
within plant tissues via the different carboxylation pathways in C3, C4 and Crassulacean acid
metabolism (CAM) plants is not relevant to understanding the mechanisms of CO2 production
explored in this study, so is not discussed here. However, the fractionations that occur during
photosynthesis ultimately dictate that C3 plants are depleted in 13C relative to atmospheric CO2, and
12
have average δ13C values of approximately −27‰ (O'Leary, 1988). C4 plants, which do not
discriminate so strongly against 13CO2 during photosynthesis, have average δ13C values around −13‰
(O'Leary, 1988). CAM plants, which utilise C3 pathways during the daytime and C4 pathways at night,
have δ13C values intermediate between C3 and C4 plants, generally between −10 and −20‰ (O'Leary,
1988).
Within plants, δ13C values differ slightly between plant roots and shoot tissues (Werth and Kuzyakov,
2010), hence different organic matter fractions have slight differences in their C isotopic
composition. Preferential utilisation of 13C-rich substrates by microorganisms, particularly in C3 soils,
acts to enrich the microbial biomass, which respires CO2 depleted in 13C relative to itself, but still
enriched compared to soil organic matter (Werth and Kuzyakov, 2010). Numerous studies have
reported various fractionations that either concur or disagree with this, yet others have shown no
indication of isotopic fractionation during the decomposition of various substrates (Boström et al.,
2007). Werth and Kuzyakov (2010) performed a comprehensive analysis of measured fractionations
between soil organic matter and soil C pools, incorporating data from field and laboratory studies
carried out at various sites around the world. Their analysis (for C3 soils) yielded a mean fractionation
between soil organic matter and soil microbial biomass of 1.2 ± 2.6‰, and a mean fractionation
between soil organic matter and soil organic matter-derived CO2 of 0.7 ± 2.8‰. These fractionations
can be considered relatively minor in terms of their effect on soil-respired δ13CCO2 values.
In arid environments, the fractionations associated with biological utilisation of organic matter and
respiration of CO2 are likely to be even less significant. Firstly, the activity of organisms is restricted to
periods when abiotic factors (primarily soil temperature and moisture content) are suitable.
Secondly, such environments tend to be characterised by low levels of soil organic matter: limited
substrate availability restricts the potential for preferential decomposition of particular soil organic
matter fractions. Therefore, the low substrate availability and abiotic factors which regulate soil
microbial activity dictate that organisms will consume all types of soil organic matter available
(Werth and Kuzyakov, 2010), and consequently, the isotopic composition of soil-respired CO2 will be
very close to that of the bulk soil organic matter pool (Dörr and Münnich, 1980).
2.2.4 Dissolution of CO2 in water
The partitioning of CO2 between gaseous and dissolved forms is determined by Henry’s Law and the
pH of the solution into which it dissolves. The Henry coefficient (KH) expresses the relationship
between CO2 in the atmosphere (pCO2) and dissolved gaseous CO2 ([CO2(aq)] = [H2CO3]) as pCO2 =
KH[CO2(aq)]. Total dissolved CO2 (CT) comprises the species H2CO3, HCO3− and CO32−, as determined by
the equilibrium:
13
𝐶𝐶𝑂𝑂2(𝑔𝑔) + 𝐻𝐻2 𝑂𝑂(𝑙𝑙) ↔ 𝐻𝐻2 𝐶𝐶𝑂𝑂3(𝑎𝑎𝑎𝑎) ↔ 𝐻𝐻 + (𝑎𝑎𝑎𝑎) + 𝐻𝐻𝐻𝐻𝑂𝑂3− (𝑎𝑎𝑎𝑎) ↔ 2𝐻𝐻 + (𝑎𝑎𝑎𝑎) + 𝐶𝐶𝑂𝑂32− (𝑎𝑎𝑎𝑎) .
(2.3)
Therefore, the relationship between CT and dissolved CO2 is:
𝐶𝐶𝑇𝑇 = �𝐶𝐶𝑂𝑂2(𝑎𝑎𝑎𝑎) � �1 +
𝐾𝐾𝑎𝑎1
𝐾𝐾𝑎𝑎 𝐾𝐾𝑎𝑎
+ 1+ 22 �.
+
[𝐻𝐻 ] [𝐻𝐻 ]
(2.4)
The equilibrium constants for [H+] [HCO3−]/[H2CO3] and [H+] [CO32−]/[HCO3−] are Ka1 and Ka2,
respectively (Stumm and Morgan, 1996).
The Henry coefficient is temperature dependent. As temperature increases, the solubility of CO2
decreases, thus pCO2 increases and [CO2(aq)] decreases. The total amount of dissolved inorganic C is
related to pH, according to equation (2.4); as pH increases ([H+] decreases), the total amount of
dissolved inorganic C (CT) increases. The relationship between pH and the dissolved CO2 species
present is expressed in Figure 2.1.
Figure 2.1
The effect of pH on the distribution of dissolved inorganic carbon species in solution.
Mechanism for abiotic CO2 fluxes
In soils, Henry’s Law-controlled dissolution and exsolution of CO2 provides a mechanism by which
abiotic CO2 fluxes can occur, with diel variability in soil temperatures driving the dissolution and
exsolution of CO2.
14
2.2.5 Carbon isotopic fractionation in the carbonate system
Isotopic fractionation of C in the carbonate equilibrium system described above enables abiotic
influences on soil CO2 fluxes to be quantified. The values for the fractionations that occur as a result
of dissolution of CO2 in water are solely temperature dependent, and are derived empirically based
on the dissolution of CO2 and subsequent dissociation reactions within equation (2.3).
As shown in Figure 2.1, pH dictates which species are present. Equilibrium fractionation of CO2 gas to
HCO3−(aq), CO32−(aq) and solid calcium carbonate as a result of CO2 dissolution results in depletion of
CO2 gas relative to HCO3−(aq), CO32−(aq), and solid calcium carbonate (Table 2.1; Thode et al., 1965;
Mook et al., 1974; Mook, 2000). At pH values <8.3, dissolution of CO2 will produce some H2CO3,
which results in enrichment of CO2 gas relative to H2CO3(aq) (Vogel et al., 1970). Therefore, dissolution
of CO2 in soil water to form HCO3− and CO32− (at pH values >8.3) would result in reduced soil CO2
concentrations and depletion of the δ13CCO2 of the soil atmosphere. Conversely, exsolution of CO2
would increase soil CO2 concentrations and lead to enrichment of the δ13CCO2 of the soil atmosphere.
This constitutes an important abiotic mechanism of CO2 uptake and release, driven by temperature,
which can occur in soils.
Table 2.1
Carbon isotope fractionation in the CO2−HCO3−CO3−CaCO3 system under equilibrium
conditions. 13εy/x represents the fractionation of compound y relative to compound x.
Soil temperatures (T) shown are similar to those measured in the McMurdo Dry Valleys.
g = gaseous CO2, a = dissolved CO2, b = dissolved HCO3−, c = dissolved CO32− ions, s = solid
calcium carbonate. TK = T (°C) + 273.15 K.
ε g/b b
(‰)
−10.70
13
1
εg/a a
(‰)
1.17
ε g/c c
(‰)
−10.06
ε g/s d
(‰)
−10.47
5
1.15
−10.20
−9.61
−10.19
T (°C)
13
13
13
a 13
εg/a = −0.19‰ + 373/TK (Vogel et al., 1970).
εg/b = 23.89‰ − 9483/TK (Mook et al., 1974).
c 13
εg/c = 13εg/b + 13εb/c (= −2.52 ‰ + 867/TK; Thode et al., 1965).
d 13
εg/s = 9.15‰ − 5380/TK (Mook, 2000).
b 13
15
2.3 McMurdo Dry Valleys
The McMurdo Dry Valleys are the largest ice-free region in Antarctica (4500 km2; Levy, 2013) and
comprise a polar desert landscape dominated by coarse-textured soils formed in glacial drift,
interspersed with alpine glaciers, ephemeral streams, and perennially ice-covered lakes (Figure 2.2).
Vascular vegetation is absent, and biota consists of microorganisms and a limited number of
invertebrate species (Adams et al., 2006; Cary et al., 2010).
Figure 2.2
The polar desert landscape of the McMurdo Dry Valleys. Clockwise from top: alpine
glaciers and ice-covered Lake Bonney, central Taylor Valley; polygonal patterned ground,
lower Taylor Valley; ephemeral stream, lower Taylor Valley.
The Dry Valleys are predominantly comprised of three large east-west oriented valleys: Taylor Valley,
Wright Valley and the Victoria Valley system. They extend from the crest of the Transantarctic
Mountains to the ocean (McMurdo Sound) and range in length from 80−100 km long, and 5−10 km
wide. Elevation ranges from sea level to over 2000 m in a horizontal distance of only a few kilometres
(Fountain et al., 1999; Doran et al., 2002a). The ice-free nature of the valleys results from the
16
Transantarctic Mountains blocking much of the flow of the East Antarctic Ice Sheet toward McMurdo
Sound (Clow et al., 1988). Additionally, low precipitation and high sublimation rates dictate that
annual ablation rates exceed accumulation rates (Fountain et al., 1999).
The Dry Valleys are one of the most intensively studied areas in Antarctica; their relative ease of
access, glacial history, and relatively simple ecosystems make them ideal for studying fundamental
landscape and ecosystem processes in a polar desert environment (Wall and Virginia, 1999). The
study area this thesis focuses on is located within Taylor Valley, the smallest and southernmost of the
three main dry valleys. Therefore, this review focuses on the climate, hydrology, geomorphology,
soils, CO2 fluxes and ecosystem processes in Taylor Valley.
2.3.1 Climate
The McMurdo Dry Valleys are a polar desert environment, characterised by a paucity of liquid water
and extreme cold conditions (Fountain et al., 1999). The austral summer provides a short opportunity
for life to flourish: daylight is continuous, temperatures are relatively mild (and are predominantly
controlled by diel variation in solar radiation (Clow et al., 1988)), and, for short periods, liquid water
is present. In contrast, sunlight and liquid water are absent during the winter months (Fountain et al.,
1999), and temperatures are controlled by the wind (Clow et al., 1988). The presence of ice within
the valleys, which exists as alpine glaciers flowing down the valley walls, ice-covered lakes, and icecemented permafrost within the soils, provides a significant pool of water that is potentially available
for melting. As such, the Dry Valleys are highly climate-sensitive environments, with small changes in
summer temperature and the associated phase changes of H2O between ice and liquid water
potentially leading to extreme variations in the hydrological regime (Fountain et al., 1999; Lyons et
al., 2005), with concomitant effects on soil chemistry, and moisture availability for organisms.
Annual water-equivalent precipitation is < 50 mm, with snow transported by katabatic winds
supplying up to another 50 mm (water-equivalent) to the valley floor (Fountain et al., 2010).
Meltwater from alpine glaciers is the dominant source of moisture for streams (Doran et al., 2002b),
as in the valley floors, ablation of snow and ice exceeds accumulation throughout the year (Clow et
al., 1988). The annual net radiation balance is positive (Dana et al., 1998) due to the relatively low
surface albedo (Thompson et al., 1971). Snow patches are generally rapidly sublimed, and hence lost
as a potential source of moisture to soils or streams (Keys, 1980).
Mean annual air temperature in Taylor Valley ranges from –14.8 to –23.1°C (Doran et al., 2002a). A
strong east to west gradient in summer temperature, which is correlated with the wind regime,
exists within the valleys; temperatures increase with distance inland (Doran et al., 2002a). In
summer, the most frequent wind direction in the valley floors is up-valley winds (easterly sea
17
breezes; Thompson et al., 1971; Keys, 1980; Clow et al., 1988; Doran et al., 2002a; Nylen et al., 2004),
generated as a result of differential heating of the dark coloured valley floors and the ice-covered
ocean. These winds typically intensify during the day (Clow et al., 1988), becoming strongest in the
afternoons and evenings when ground temperatures are warmest (Keys, 1980). The easterly winds
also warm as they progress up-valley over the relatively warmer dark valley surfaces; air
temperatures increase by 0.09°C per kilometre inland and cool with inland elevation at the dry
adiabatic lapse rate (Doran et al., 2002a).
Westerly katabatic winds, which warm adiabatically as they descend into the valleys (Keys, 1980;
Clow et al., 1988), occur more frequently in winter than in summer (Clow et al., 1988; Nylen et al.,
2004). In summer, they occur infrequently (Nylen et al., 2004), and are generally limited to the
mornings, when solar radiation is minimal, thus the atmosphere is more stable (Thompson et al.,
1971; Clow et al., 1988). The onset and cessation of katabatic winds is generally abrupt, though the
adiabatic warming of the winds, combined with the disruption they cause to temperature inversions
(Thompson et al., 1971; Doran et al., 2002a; Nylen et al., 2004) acts to increase air temperatures and
decrease relative humidity (Thompson et al., 1971; Clow et al., 1988; Nylen et al., 2004). Katabatics
can reach speeds of 37 m s-1 and can increase winter air temperatures by 30°C in a few hours (Doran
et al., 2002a; Nylen et al., 2004), with the warming effect of the katabatics persisting for days after
the winds cease (Nylen et al., 2004).
Mean annual solar flux in the Dry Valleys ranges from 73 – 117 W m-2 (Doran et al., 2002a). Strong
diel variability in solar radiation, further influenced by topographic shading and localised cloudiness,
acts to drive diel variation in air and soil temperatures, glacial meltwater production, and stream
flow.
2.3.2 Hydrology
Small changes in summer temperature and solar radiation affects the availability and distribution of
liquid water, which has a significant influence on ecological function and biological diversity in the
Dry Valleys (Doran et al., 2002b). Taylor Valley contains three major lakes, and many ephemeral
streams and ponds (Fountain et al., 1999; Moorhead et al., 2003; Moorhead, 2007). The major lakes
are perennially ice-covered, although absorption of radiation during summer causes melting at the
ice-margins, forming moats around the edges of the lakes (Hendy, 2000). These moats and other icefree ponds are zones of biological productivity during most summers; algae produced within the
lakes and ponds represent a significant contemporary organic C source in the landscape (Moorhead
et al., 2003; Moorhead, 2007).
18
Ephemeral streams, fed by glacial melt, provide a significant source of liquid water to hydrological
and biological systems within the Dry Valleys. Such streams generally flow for between four and ten
weeks during the austral summer (Fountain et al., 1999; McKnight et al., 1999), with flow rates
markedly influenced by diel variation in solar radiation. Water tracks also provide an extensive
topographically-controlled transport network that distributes meltwater from snow patches and
melting of ice-cement through soils, and ultimately to lakes (Levy et al., 2011).
The variable nature of solar radiation received in the Dry Valleys produces strong daily and seasonal
variations in glacier melting and stream flow (Conovitz et al., 1998; Dana et al., 1998). Hydrologic
data from both Taylor and Wright Valleys highlights the significant variability in stream flow over
daily, seasonal, and interannual time scales (Conovitz et al., 1998; McKnight et al., 1999; Doran et al.,
2008). Such variation is predominantly a result of differences in solar intensity on source glaciers,
which is primarily influenced by solar geometry and the aspect of the glaciers, but also by altitude, air
temperature (Clow et al., 1988), and cloud cover (Conovitz et al., 1998; Dana et al., 1998). However,
Doran et al.’s (2008) comparison of high and low stream-flow summers showed that despite solar
radiation being the same for the two summers compared, increased air temperatures in the highflow year, driven by an increased frequency of westerly winds, was the primary reason for increased
meltwater production.
Discrete climate events such as the relatively warm summer of 2001/2002 can also have a significant
influence on Dry Valley hydrology (Barrett et al., 2008b; Doran et al., 2008), acting to increase
stream-flow and lake levels, and melting of ice-cement within soils. Melting of subsurface ice during
warm summers can lead to the development of groundwater seeps (Lyons et al., 2005), and can
influence water availability in surface soils for several years, even if temperatures are average or
below-average in subsequent seasons (Barrett et al., 2008b). Furthermore, increased moisture
supply to soils can significantly decrease soil salinity, thereby improving habitat suitability and
altering soil faunal community abundance and diversity (Nielsen et al., 2012).
2.3.3 Geomorphology
The McMurdo Dry Valleys are east-west oriented deglaciated valleys. In Taylor Valley, the valley
floors are covered by tills of various thicknesses, preserving evidence of ice advances from Taylor
Glacier (draining the East Antarctic Ice Sheet), the Ross Sea, and local alpine glaciers. The eastern
part of Taylor Valley is mantled with Ross Sea drift, deposited when a thick lobe of westward-flowing
ice, grounded in McMurdo Sound, blocked the valley mouth and dammed Glacial Lake Washburn
(Hall et al., 2000). The drift is predominantly comprised of sediments which were deposited via a
lake-ice conveyor operating on Glacial Lake Washburn, a large proglacial lake which filled central and
lower Taylor Valley to around 300 m elevation from ~12.4−23.8 ka BP (Denton et al., 1989).
19
Consequently, the valley floors contain lacustrine sediments, and lacustrine organic material provides
evidence of shorelines delineating former lake levels. Ross Sea drift is also characterised by the
presence of kenyte erratics, sourced from Ross Island (Hall et al., 2000).
In the central and upper parts of the valley, advances of Taylor Glacier are recorded by at least four
drift sheets. Tills are mainly comprised of granite, granodiortie, diorite, gneiss and schist of the
Precambrian Ross System, with small amounts of diabase and sandstone from the Beacon
Supergroup (Pastor and Bockheim, 1980). Taylor IV drift contains several units of different age,
representing at least four expansions of Taylor Glacier (Wilch et al., 1993). However, these units can
be difficult to distinguish, and are referred to only as Taylor IVa and Taylor IVb (Bockheim et al.,
2008), with ages of <1.50 Ma and >3.47 Ma, respectively (Wilch et al., 1993). Taylor III drift is dated
as 250−340 ka, and contains reworked lacustrine carbonates (Higgins et al., 2000b). Taylor II/Bonney
drift (70−130 ka; Higgins et al., 2000b) extends up to around 300 m elevation on the valley walls
between Lake Bonney and Canada Glacier, and is dominantly comprised of yellowish silty meltout till
and gravelly sandy waterlain till (Higgins et al., 2000a). Bonney drift also contains lacustrine
carbonates, which were deposited in a proglacial lake of Taylor Glacier (Hendy et al., 1979; Higgins et
al., 2000b). Moraines associated with the present advance of Taylor Glacier, which flows into the
western part of the valley and is currently in its most extensive Holocene position, are referred to as
Taylor I (Higgins et al., 2000a). The advances of alpine glaciers in Taylor Valley are in-phase with
advances of Taylor Glacier (Hendy et al., 1979; Higgins et al., 2000a); these advances are out-ofphase with westward ice-expansion up-valley from grounded ice in the Ross Sea (Hendy et al., 1979).
2.3.4 Soils
The soil-forming environment of the McMurdo Dry Valleys is characterised by hyper-arid climatic
conditions, reflecting the prevailing low temperatures, low levels of precipitation, and low humidity
within the valleys. The prevailing cold temperatures act to restrict the availability of water in the
landscape, and dictate that, outside the short period of relative warmth in December and January,
any water present is generally frozen. However, individual site factors such as aspect, parent material
(through its influence on surface albedo) and proximity to glaciers or meltwater channels are also
important in determining soil moisture availability. Soils developed on northerly aspects are
considerably warmer than those on southerly aspects, and soils formed from darker coloured parent
materials absorb more heat from incoming radiation than those formed in lighter coloured parent
materials (Campbell and Claridge, 2006).
20
Classification
Soils of the McMurdo Dry Valleys are classified in the Gelisol soil order of Soil Taxonomy (Bockheim,
2002), with further subdivision based on soil climatic regimes and soil properties. Soils with icecemented permafrost within the upper 70 cm of the profile are generally cryoturbated, and are
classified in the Turbel suborder, whereas soils with dry-frozen permafrost tend to have little or no
evidence of cryoturbation, and are classified as Orthels (Bockheim and McLeod, 2008; Bockheim et
al., 2008). The anhydrous conditions (< 50 mm annual water equivalent precipitation) prevalent
throughout the Dry Valleys dictates that most soils are in Anhy- great groups; soils adjacent to lakes
or meltwater streams, or in moist coastal areas are classified in Haplo- great groups.
Soils in the eastern part of Taylor Valley are generally ice-cemented, and are mapped as Typic
Haploturbels (Bockheim and McLeod, 2008; Bockheim et al., 2008). In central and upper Taylor
Valley, Typic Anhyorthels (containing dry-frozen permafrost) are dominant on Taylor III drift, whereas
Salic and Petrosalic Anhyorthels occur on older surfaces (Bockheim and McLeod, 2008; Bockheim et
al., 2008), reflecting increased salt accumulation with soil age.
Soil morphology
Soils within the McMurdo Dry Valleys are typically overlain by a thin, gravelly desert pavement. The
desert pavement often contains varnished and ventifacted clasts, with the proportion of each
increasing with surface age (Bockheim, 2010). Beneath the desert pavement, soils sometimes contain
a horizon with vesicular porosity, which is dominated by sand and silt-sized particles that are thought
to be produced by weathering, rather than via dust deposition (Bockheim, 2010). These surficial
layers typically overlie gravelly sands, with the level of oxidation generally reflecting surface age
(Campbell and Claridge, 1982). Soils in the Dry Valleys contain permafrost, which is ice-cemented
where there is sufficient moisture, or “dry frozen” in the absence of sufficient moisture to cause
cementation (Bockheim, 1980; Bockheim, 1997; Campbell et al., 1998). The spatial distribution of icecemented soils varies with soil age, elevation, distance inland, climatic conditions, and proximity to
potential moisture sources (Campbell and Claridge, 2006). Characteristic changes in soil morphology
and surface weathering features with increasing soil age are shown in Table 2.2.
21
Table 2.2
Weathering
stage
Changes in soil morphology and surface weathering features with increasing weathering
(Campbell and Claridge, 1975).
Surface rock
characteristics
Soil colour
Horizon
development
Salt
morphology
Soil depth
1
fresh, unstained, coarse
& angular
pale olive to
light grey (5Y
6/3 - 7/2)
nil
absent
very shallow,
underlain by
ice
2
light staining, slight
rounding, some
disintegration
pale brown to
light brownish
grey (10YR
6/3-2.5Y 6/2)
weak
few flecks
shallow,
underlain by
ice
3
distinct polish, staining &
rounding, some
cavernous weathering,
some ventifacts
light yellowish
brown (10 YR
5/3-2.5Y 6/4)
distinct
many flecks in
upper part of
profile &
beneath stones
moderately
deep
4
boulders much reduced
by rounding, crumbling
& ventifaction, strongly
developed cavernous
weathering; staining &
polish well developed;
some desert varnish
yellowish
brown (10YR
5/4) in upper
horizons,
paler in lower
horizons
very distinct
in continuous or
discontinuous
horizon beneath
surface
deep
5
few boulders, many
pebbles forming
pavement, extensive
crumbling, staining,
rounding, pitting &
polish
dark yellowish
brown to
yelolowish red
(10YR 4/4-5YR
5/8)
very distinct
in horizon 20-30
cm beneath
surface &
scattered
throughout
profile
deep
6
weathered & crumbled
bedrock, very strongly
stained
strong brown
to yellowish
red & dark red
(7.5YR 5/65YR 4/8 or
2.5YR 3/6)
very distinct
(as for Stage 5)
shallow to
deep
Salts are a prominent feature of Antarctic soils (Claridge and Campbell, 1977; Keys and Williams,
1981; Bockheim, 1997), and occur as encrustations beneath surface clasts, disseminated salt flecks
within the matrix, or salt pans (Campbell and Claridge, 1982). A morphogenetic sequence of soluble
salt accumulation in cold desert soils has been defined (Bockheim, 1990; Table 2.3). This sequence
enables comparisons of relative age to be made between soils with varying salt accumulations,
although the influence of local climate must be carefully considered, as salts can be readily leached in
areas with greater moisture availability (Toner et al., 2013). Nonetheless, salt encrustations are more
common in older soils, evident as indurated pans at their maximum expression (Campbell and
Claridge, 1975; Bockheim, 1990).
22
Table 2.3
Morphogenetic stages of accumulation of soluble salts in Cold Desert soils (Bockheim,
1990).
Salt stage
Maximum salt
morphology
Electrical conductivity
(dS/m)
Numerical age
0
None
<0.6
<10 ka
I
Coatings on stone
bottoms
0.6-5.0
10−18 ka
II
Few flecks (<20% of
surface area has flecks
1 – 2 mm in diameter)
5−18
18−90 ka
III
Many flecks (>20% of
surface area has flecks
1 – 2 mm in diameter)
18−25
90−200 ka
IV
Weakly cemented pan
25−40
250−? ka
V
Strongly cemented
pan
40−60
~ 1.7−2.5 Ma
VI
Indurated pan
60−100+
~ >2.5 Ma
Chemical properties
Chemical weathering processes are limited by the cold, dry climatic conditions, although oxidation of
parent materials and precipitation of secondary minerals does occur (Claridge, 1965; Campbell and
Claridge, 1975; Keys and Williams, 1981; Bockheim, 1997). The dominant soil-forming processes
occurring are salinisation and oxidation (Bockheim, 1990).
Soil pH values range from neutral to alkaline (Bockheim, 1980), depending on the balance between
acidic salts derived from atmospheric deposition, and alkaline salts derived from local marine inputs
(Claridge and Campbell, 1977). Consequently, trends in soil pH are closely related to the distribution
of salts. Higher pH values are measured in coastal zones where marine influxes are dominant,
whereas lower pH values measured inland reflect the influence of atmospheric deposition on soil salt
composition (Claridge and Campbell, 1977; Keys and Williams, 1981). Electrical conductivity, which is
a proxy for saltiness, varies according to age and soil moisture regime. The dominant cation in 1:5
soil:water extracts of soils older than 18 ka is sodium (Bockheim, 1980); this further highlights the
23
arid nature of the soil-forming environment, with the low levels of soil moisture effectively
preserving the presence of soluble salts.
Soluble salts
Salt input and migration can be considered one of the few active and enduring processes occurring in
the hyperarid environment of the Dry Valleys (Bao et al., 2008). The stability of the landscape over
long timescales means that analysis of salts within the soils provides opportunities to investigate
processes controlling the distribution of salts in the landscape, as well as the sources of salts and
their post-depositional transformations and translocations.
Salts present within soils of the McMurdo Dry Valleys are predominantly chlorides, nitrates and
sulfates of sodium, potassium, calcium and magnesium (Claridge and Campbell, 1977; Keys and
Williams, 1981). Early studies of the origin of these salts concluded that calcium, magnesium and
potassium are largely derived from rock weathering, and sodium is predominantly of marine origin.
Chlorides, nitrates and sulfates were also considered to be of marine origin, with nitrate and sulfate
predominantly derived from atmospheric circulation, and chloride from local air masses (Claridge and
Campbell, 1977).
More recent studies (e.g. Michalski et al., 2005; Bao and Marchant, 2006; Bao et al., 2008) have
utilised stable isotope ratios of anions to quantitatively determine their origins. The oxygen isotopic
composition of nitrates in soils of the Dry Valleys indicates that their source is solely atmospheric in
origin, comprising roughly equal proportions of atmospheric nitrate from the troposphere, and
stratospheric HNO3 (Michalski et al., 2005). These salts are transported to Dry Valley soils either by
oceanic air masses moving up-valley, or by down-valley katabatic winds picking up nitrates deposited
on the polar plateau (Michalski et al., 2005). Given the general trend of increasing nitrate
concentrations with distance inland and soil age (Claridge and Campbell, 1977; Keys and Williams,
1981; Bockheim, 1997), it appears that transport via katabatic winds off the polar plateau accounts
for the majority of nitrate deposition in the Dry Valleys.
A study of the isotopic composition of sulfates in soils of the McMurdo Dry Valleys revealed three
components contribute to the total sulfate content of soils. Sea-salt sulfate and background sulfate,
derived from weathering and volcanic sources, are more abundant in coastal zones, whereas nonsea-salt sulfate, which predominantly consists of photochemically generated sulfate derived from the
oxidation of dimethylsulfide, dominates in soils of the upland valleys (Bao and Marchant, 2006).
By investigating the isotopic composition of chloride, Bao et al. (2008) identified two major sources
of chloride: airborne sea-salt chloride, and secondary atmospheric chloride. They also found that
chloride distribution within the Dry Valleys is explained by the balance between continuous influx
24
from atmospheric deposition and variable outflux through leaching. Chloride influx decreases inland,
as the dominant sea-salt component is transport-limited (Claridge and Campbell, 1977; Keys and
Williams, 1981; Bao et al., 2008). Similarly, chloride outflux decreases inland, as soil temperatures
decrease with increasing distance (and elevation) from the coast, thereby limiting moisture
availability and thus potential for leaching. Interestingly, chloride concentrations were relatively low
in both the coastal thaw and stable upland zones, and highest (yet highly variable) in the inland
mixed zone (Bao et al., 2008). This distribution can be explained by relatively high rates of leaching in
the moist coastal thaw zone, and relatively low inputs of chloride in the stable upland zone. The high
concentrations measured in the inland mixed zone show that the rate of chloride influx exceeds the
rate of outflux via leaching.
The leaching of chloride in relatively moist coastal zones enables the typically high soil pH values
measured in coastal areas to be explained. So long as liquid moisture is present, there is effectively
an unlimited supply of HCO3− via dissolution of CO2 in soil solution. Na+ from local sea-salt aerosolic
deposition is relatively abundant, hence NaHCO3− is readily formed. This reaction displaces calcium,
available in relatively small quantities via rock weathering (Claridge and Campbell, 1977; Keys and
Williams, 1981), calcite dissolution (Toner et al., 2013) or aerosolic dust inputs (Foley et al., 2006),
which is likely to be leached with chloride ions as Ca-Mg-Cl brines (Toner et al., 2013). The presence
of NaHCO3− acts to increase soil pH to values above 8.34 (the pH that can be generated by the
presence of calcium carbonate alone). In soils of the stable upland zone, chloride concentrations are
low, moisture is generally absent, and consequently, soil pH values are lower.
Carbon
Soil organic C contents are very low, typically ranging between 0.01 and 0.05% (Burkins et al., 2001).
Inorganic C concentrations are an order of magnitude greater than organic C concentrations, ranging
from 0.02 to 0.3% (Burkins et al., 2001; Foley et al., 2006), thus constituting the largest terrestrial C
pool in ice-free areas of Antarctica.
Two different models of C cycling have been proposed for the McMurdo Dry Valleys. The “legacy
model” suggests that organic matter present at lower elevations in the Dry Valleys is a legacy of
ancient glacial tills and lacustrine systems, and organic matter at higher elevations is a result of longterm in situ autotrophic activity in soils (Burkins et al., 2000). An alternative “spatial subsidies model”
proposes that present lakes provide the dominant source of organic matter which is subsequently
distributed around the valleys by wind (Elberling et al., 2006). This model, however, acknowledges
that the two hypotheses are not mutually exclusive, with C present in Dry Valley soils perhaps being
partially sourced from ancient lakes as well as receiving a modern subsidy, with the relative
25
proportions from each source likely influenced by valley size and the proportion of the valley
occupied by lakes (Elberling et al., 2006).
Attempts to constrain the relative importance of legacy effects on C cycling in Dry Valley ecosystems
have been impeded by the lack of conspicuous sources of legacy organic matter (Hopkins et al.,
2009). As such, the isotopic composition of soil organic matter continues to provide a useful
indication as to its origin, but cannot definitively provide an indication as to its relative age. Soils on
valley floors are predominantly characterised by lacustrine-derived organic matter, reflecting either
the influence of large paleolakes or contemporary lakes within the valleys, whereas soils at higher
elevations tend to have organic matter signatures characteristic of endolith-derived organic matter
(Burkins et al., 2000; Hopkins et al., 2009).
Current estimates of C turnover times in the McMurdo Dry Valleys derived from in situ soil CO2 flux
rates are between 23 and 123 years (Burkins et al., 2001; Elberling et al., 2006). Such rapid turnover
times suggest that there would be no legacy C remaining in the soils. However, these estimates are
based on measured CO2 effluxes which are assumed to be biologically driven.
Biota
Despite climatic conditions appearing unfavourable for life, diverse biological communities exist in
the lakes, streams and soils (Fountain et al., 1999). Such communities are adapted for survival in
biologically extreme conditions including freeze-thaw cycles, high winds (Fountain et al., 1999),
prolonged low temperatures, hypersalinity, total darkness during winter and exposure to high
intensities of solar radiation during summer, and low levels of moisture and organic C.
Soil organisms in the Dry Valleys are limited to microorganisms and a few invertebrate species
(nematodes, rotifers, tardigrades, springtails and mites; Adams et al., 2006). Springtails and mites are
the largest terrestrial invertebrates, but nematodes are the most abundant invertebrates in the Dry
Valleys (Freckman and Virginia, 1997; Treonis et al., 1999; Treonis and Wall, 2005). The distribution
of nematodes across a wide range of soil habitats is enhanced by their ability to undergo
anhydrobiosis, a survival strategy induced by desiccation (Treonis et al., 2000; Treonis and Wall,
2005).
The low-complexity ecosystems of the Dry Valleys provide opportunities to investigate factors
controlling ecosystem structure and function in the absence of vascular plants. Recent studies have
shown that the distribution and diversity of biota within Dry Valley terrestrial ecosystems is largely
controlled by abiotic factors, with soil moisture content and salinity identified as key factors over a
range of sites (Barrett et al., 2004; Smith et al., 2010; Lee et al., 2012; Magalhães et al., 2012; Stomeo
et al., 2012; Van Horn et al., 2013).
26
2.3.5 CO2 fluxes
Respiration
Previous studies of soil respiration in Dry Valley ecosystems have been conducted in order to
quantify the turnover time of the soil organic C pool (Burkins et al., 2001; Barrett et al., 2006a;
Elberling et al., 2006), and to provide a basis for comparing ecosystem functioning between sites
(Barrett et al., 2006b). Surface CO2 fluxes in the McMurdo Dry Valleys generally range from −0.1 to
0.15 µmol CO2 m-2 s-1 (e.g. Burkins et al., 2001; Parsons et al., 2004; Elberling et al., 2006; Ball et al.,
2009), although at sites with higher C and moisture contents, rates can be as high as 0.78 µmol CO2
m-2 s-1 (Gregorich et al., 2006). Previous studies have generally focused on relating soil respiration
rates to environmental variables such as soil temperature and moisture content (Parsons et al., 2004;
Elberling et al., 2006; Gregorich et al., 2006) and observing responses to in situ (Burkins et al., 2001;
Ball et al., 2009) and laboratory manipulations (Burkins et al., 2001; Treonis et al., 2002; Parsons et
al., 2004; Barrett et al., 2005, 2006a; Elberling et al., 2006; Hopkins et al., 2006; Hopkins et al., 2008;
Hopkins et al., 2009).
Studies of Dry Valley soil respiration have generally been predicated on the basis that, as in
temperate environments, CO2 fluxes are biologically produced, and the maximum rate of CO2
production occurs during the warmest part of the day. However, as the diversity, abundance and
activity of biota that inhabit Dry Valley soil ecosystems is comparatively low, and largely controlled
by abiotic factors (Barrett et al., 2004; Lee et al., 2012; Magalhães et al., 2012; Stomeo et al., 2012;
Van Horn et al., 2013), this assumption may not necessarily be valid, and therefore needs to be
tested. Additionally, none of the previous measurements of soil CO2 flux in the McMurdo Dry Valleys
have identified the depths at which CO2 production occurs, and have therefore not considered
potential lags between CO2 production rate and surface CO2 flux.
However, studies of soil CO2 flux rates over several diel cycles by Parsons et al. (2004) and Ball et al.
(2009) found that soil CO2 fluxes in the McMurdo Dry Valleys follow diel variations, with changes in
soil temperature accounting for almost half of the variation in CO2 flux (Ball et al., 2009). Both
authors conclude that soil CO2 flux is strongly influenced by abiotic factors, and suggest that
dissolution of CO2 in water, consistent with Henry’s Law, is the likely explanation for the diel variation
observed.
Discriminating between biotic and abiotic processes using stable carbon isotopes
An abiotic contribution to soil CO2 fluxes has potentially profound implications for interpretation of
the fluxes in terms of soil respiration and C cycling in the McMurdo Dry Valleys. Investigating the
isotopic composition of surface CO2 fluxes and subsurface soil CO2 profiles is one way of determining
the extent to which Henry’s Law-controlled dissolution and exsolution of CO2 can account for Parsons
27
et al. (2004) and Ball et al.’s (2009) observed diel variation in surface CO2 flux rates. By using a mixing
model that incorporates the isotopic composition of two end members, namely biotically and
abiotically-produced CO2, the proportion of biotic and abiotic processes that contribute to measured
surface CO2 fluxes can be quantified.
Natural abundance isotope studies of soils from the McMurdo Dry Valleys have been used to trace
the provenance of organic material within the soils (e.g. Burkins et al., 2000; Hopkins et al., 2009),
and in laboratory studies of organic matter decomposition (Hopkins et al., 2009), but no field studies
have incorporated isotopic analyses of soil CO2 samples. The only profile gas concentrations
measured in the McMurdo Dry Valleys are from Gregorich et al.’s (2006) study of relatively moist,
organic-rich lakeshore soils in Garwood Valley, where high levels of CO2 (up to 5550 µL L-1) were
measured at 5 and 10 cm depth. The isotopic composition of the CO2 was not determined.
δ13C values associated with biotic CO2 production and consumption
Biotic production and consumption of CO2 and the abiotic dissolution/exsolution processes outlined
above produce distinctive δ13CCO2 values. Biological respiration produces δ13CCO2 values that reflect
the δ13C of the source organic material (Dörr and Münnich, 1980), as respiration does not (or only to
a minor extent) discriminate against 13C. In soils of the McMurdo Dry Valleys, sources of soil organic C
have values that range between −10 and −19‰ for lacustrine detritus and from −22 to −30.5‰ for
moss and endolith-derived organic matter (Burkins et al., 2000; Hopkins et al., 2009). Any CO2
produced by biological respiration would have a similarly depleted δ13CCO2 value. In contrast,
consumption of CO2 via algal photosynthesis would enrich soil δ13CCO2 values, as preferential uptake
of 12CO2 by photosynthetic organisms acts to concentrate the heavier 13CO2 in the soil atmosphere
(Park and Epstein, 1960). However, given that photosynthesis, by definition, requires light in order to
occur, such processes can be readily eliminated at depths beneath the soil surface.
δ13C values associated with abiotic CO2 production and consumption
The fractionation of soil CO2 associated with dissolution and exsolution processes suggests that any
abiotically-produced CO2 will be of substantially different isotopic composition to that of biologicallyproduced CO2. This means that a 2-component mixing model should be able to be used to determine
the relative proportions of biotic and abiotic processes contributing to soil CO2 fluxes in Dry Valley
soils. At pH values >9, dissolved inorganic C species are dominated by the HCO3− and CO32− forms
(Stumm and Morgan, 1996), and equilibrium fractionation of CO2 gas to HCO3− and CO32− results in a
~ −10‰ depletion of CO2 gas relative to HCO3− or CO32− (Thode et al., 1965; Mook et al., 1974; Table
2.1). Therefore, dissolution of CO2 in soil water to form HCO3− and CO32− (at high soil pH values and
relatively low soil temperatures typical of Dry Valley soils) would result in the depletion of δ13CCO2 in
28
the soil atmosphere. Conversely, exsolution of CO2 would lead to enrichment of δ13CCO2 in the soil
atmosphere.
2.4 Research gaps
This review has established that Dry Valley soil CO2 fluxes have the potential to be strongly
influenced by abiotic factors, with Parsons et al. (2004) and Ball et al. (2009) suggesting Henry’s Lawcontrolled dissolution and exsolution of CO2 as a driving mechanism. However, as yet there has been
no attempt to quantify the potential abiotic contribution to CO2 fluxes in Dry Valley soils, and thereby
test the assumption made in previous studies that Dry Valley soil CO2 fluxes are biologically
produced.
This study aims to characterise and quantify the detailed variations in surface CO2 flux rates and
subsurface CO2 concentration and δ13CCO2 profiles at two sites with contrasting organic C inputs. This
research extends previous work on Dry Valley soil CO2 fluxes by using high-resolution measurements
of surface CO2 flux rate and, for the first time, subsurface CO2 concentration and δ13CCO2 profiles in
order to link surface and subsurface soil CO2 dynamics, and to discriminate between biotic and
abiotic sources of CO2.
The research presented in this thesis will contribute to an improved understanding of the factors and
processes governing CO2 fluxes in Dry Valley soils by addressing the following research questions:
•
Can exsolution and dissolution of CO2 in accordance with Henry’s Law account for diel
variation in surface CO2 flux rates (Chapter 3)?
•
What are the environmental controls on biotic and abiotic CO2 production in soils with
contrasting organic C supplies (Chapters 4 and 5)?
•
What are the relative contributions of biotic and abiotic processes to soil CO2 fluxes in 1) a
dry soil with low organic C content, which is representative of “typical” Dry Valley soils
(Chapter 4), and 2) a soil with relatively high levels of organic C, which is characteristic of a
biological “hotspot” (Chapter 5)?
29
30
Chapter 3
Abiotic processes dominate CO2 fluxes in Antarctic soils
This chapter is written as a manuscript, and has been published in Soil Biology and Biochemistry. The
citation is as follows:
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., 2012. Abiotic processes dominate CO2 fluxes
in Antarctic soils. Soil Biology & Biochemistry, 53: 99−111. doi:10.1016/j.soilbio.2012.04.027.
3.1 Abstract
Ecosystems within the McMurdo Dry Valleys of Antarctica are highly sensitive to environmental
change. Increases in soil temperature and/or moisture content may dramatically change rates of soil
respiration and soil carbon (C) turnover. Present estimates of soil respiration rates and C turnover
times are based on surface carbon dioxide (CO2) fluxes and soil organic C content. However, the
assumption that surface CO2 fluxes are purely biological in origin has not been rigorously tested. This
study uses concentration and, for the first time, the stable C isotopic composition of surface soil CO2
fluxes and subsurface CO2 profiles to: 1) examine mechanisms of soil CO2 uptake and release, 2)
identify the location of potential CO2 sources and sinks within the soil profile, and 3) discriminate
between biotic and abiotic contributions to CO2 fluxes in soils of Taylor Valley. Surface CO2 fluxes and
subsurface CO2 profiles confirm that these soils take up and release CO2 on a daily basis (during the
austral summer), associated with small changes in soil temperature. Shifts in the C isotopic
composition of soil CO2 are inconsistent with biological mechanisms of CO2 production and
consumption. Instead, the isotopic shifts can be accounted for by Henry’s Law dissolution and
exsolution of CO2 into a solution of high pH, driven by changes in soil temperature. The results
constrain the biological component of soil CO2 fluxes in Taylor Valley to less than 25% (and likely to
be significantly less). This finding implies that previous measurements of surface soil CO2 fluxes are
overestimates of soil respiration, thus C turnover times calculated from them are underestimates.
Discriminating between biotic and abiotic contributions to CO2 fluxes in Antarctic Dry Valley soils is
essential if the effects of climate change on these sensitive ecosystems are to be accurately
identified.
31
3.2 Introduction
Predicting the impact of changes in climate conditions on Antarctic ice-free areas necessitates a
detailed understanding of the rates and dynamics of processes operating in these extreme
environments. Previous studies of soil respiration in Antarctic ecosystems have been conducted in
order to quantify the turnover time of the soil organic carbon (C) pool (Burkins et al., 2001; Barrett et
al., 2006a; Elberling et al., 2006), and to provide a basis for comparing ecosystem functioning
between sites (Barrett et al., 2006b). Investigations of Antarctic soil respiration rates have also
provided insight into the physical controls on soil carbon dioxide (CO2) flux (Parsons et al., 2004) and
the effects of climate variation on C cycling (Ball et al., 2009) in dry valley ecosystems. Surface CO2
fluxes in the Dry Valleys generally range from −0.1 to 0.15 µmol CO2 m-2 s-1 (e.g. Burkins et al., 2001;
Parsons et al., 2004; Elberling et al., 2006; Ball et al., 2009), although at sites with higher C and
moisture contents, rates can be as high as 0.78 µmol CO2 m-2 s-1 (Gregorich et al., 2006). Previous
studies have generally focused on relating soil respiration rates to environmental variables such as
soil temperature and moisture content (Parsons et al., 2004; Elberling et al., 2006; Gregorich et al.,
2006) and observing responses to in situ (Burkins et al., 2001; Ball et al., 2009) and laboratory
manipulations (Burkins et al., 2001; Treonis et al., 2002; Parsons et al., 2004; Barrett et al., 2005,
2006a; Elberling et al., 2006; Hopkins et al., 2006; Hopkins et al., 2008; Hopkins et al., 2009).
Studies by Parsons et al. (2004) and Ball et al. (2009) found that soil CO2 fluxes in the McMurdo Dry
Valleys follow diel variations, with soil temperature accounting for almost half of the variation in CO2
flux (Ball et al., 2009). Both authors conclude that soil CO2 flux is strongly influenced by abiotic
factors, and suggest that dissolution of CO2 in soil water, consistent with Henry’s Law, is the likely
explanation for the diel variation observed. This is in contrast to many other ecosystems, where
temperature predominantly influences biotic factors.
An abiotic contribution to soil CO2 flux has potentially profound implications for interpretation of
CO2 fluxes in terms of soil respiration and C cycling in the McMurdo Dry Valleys. Present estimates of
turnover times based on in situ soil respiration rates are between 23 – 130 years (Burkins et al., 2001;
Elberling et al., 2006). Any abiotic contributions to CO2 fluxes would mean that soil respiration rates
overestimate the biological contributions, and turnover times calculated from such fluxes would be
underestimates.
This chapter presents data from a study aimed to advance the mechanistic understanding of the
processes governing CO2 fluxes in Antarctic soils. Surface soil CO2 fluxes and subsurface soil CO2
concentration profiles were used to characterise diel soil CO2 dynamics, and identify potential CO2
sources and sinks within the soil profile. This enabled the hypothesis of Ball et al. (2009) and Parsons
et al. (2004) to be tested by: 1) comparing measured subsurface soil CO2 concentrations with values
32
simulated by a model that incorporates Henry’s Law dissolution of CO2, and 2) quantifying the biotic
and abiotic components of soil CO2 by utilising the contrasting relationships between concentration
and C isotopic composition of soil CO2 (δ13CCO2) characteristic of biotic and abiotic processes.
3.3 Materials and Methods
3.3.1 Study area
The study was conducted during the 2008/09 austral summer at two sites in Taylor Valley, in the
McMurdo Dry Valleys, Antarctica. The McMurdo Dry Valleys are the largest ice-free region in
Antarctica (4500 km2; Levy, 2013) and comprise a polar desert landscape dominated by coarsetextured soils formed in glacial drift, interspersed with alpine glaciers, ephemeral streams, and
perennially ice-covered lakes. Vascular vegetation is absent, and biota consist of microorganisms and
a limited number of invertebrate species (Adams et al., 2006; Cary et al., 2010). Mean annual air
temperature in Taylor Valley ranges from −14.8°C to −23.1°C (Doran et al., 2002a), and annual waterequivalent precipitation is < 50 mm, with snow transported by katabatic winds supplying up to
another 50 mm (water-equivalent) to the valley floor (Fountain et al., 2010).
Figure 3.1
Location of Sites A and B, Taylor Valley, Antarctica.
33
Site A was on relatively young glacial drift (Ross Sea Drift, ~ 9 − 28 ka; Hall and Denton, 2000),
adjacent to a small lake (~0.05 km2; known as Spaulding Pond) situated 700 m east of Howard Glacier
(77° 39.572’S, 163° 07.394’E; Figure 3.1). This site was selected to provide soils with relatively low
salt concentrations and higher inputs of organic C. During summer, when a moat forms around the
edge of the lake, algae are driven to the lake margins by winds and are eventually blown onshore,
providing a local source of organic C (Figure 3.2). Soils at Site A are weakly developed, strongly
cryoturbated, and contain ice-cemented permafrost within 70 cm of the soil surface (Typic
Haploturbels; Bockheim et al., 2008). At the time of sampling (late December 2008), soils at Site A
were ice-cemented below 35 cm depth.
Site B was on an older moraine crest (Taylor III Drift, 250 − 340 ka; Higgins et al., 2000b),
approximately 1.5 km west of Howard Glacier (77° 39.540’S, 163° 00.484’E). Soils at Site B were
expected to have higher salt concentrations due to their age, and lower organic C levels as there was
no lake nearby to provide a local source of organic material (Figure 3.3). Soils formed in Taylor III drift
are mapped as Typic Anhyorthels (Bockheim et al., 2008), which by definition should not contain icecemented permafrost in the upper 70 cm of the profile. However, soils at Site B had a similar
morphology to those at Site A (Figure 3.4; Typic Haploturbels), and were ice-cemented below 26 cm
depth. Both sites were characterised by patterned ground comprised of flat-centred polygons
measuring approximately 15 x 15 m.
34
Figure 3.2
Site A, lower Taylor Valley. Clockwise from top: view west to gas sampling area and
Howard Glacier; headspace chambers with Spaulding Pond in the background; algae
accumulating at the edge of Spaulding Pond.
35
Figure 3.3
Site B, lower Taylor Valley. Top: view west showing gas sampling chambers and tubes.
Bottom: view south-east showing soil pit on broad moraine crest. Site A is located next
to the lake in the distance on the left hand site of the photo.
36
Figure 3.4
Soils (Typic Haploturbels) at Site A and Site B. At the time of sampling (late December
2008), soils at Sites A and B were ice-cemented below 35 and 26 cm, respectively. The
profile shown for Site B has been deepened with a concrete breaker and is moistened
from melting of ice-cemented permafrost.
3.3.2 Soil temperature and moisture content measurements
Soil temperature and moisture content at 5, 15 and 25 cm depth were measured using Hydra Probes
(Hydra Probe II, Stevens Water Monitoring Systems Inc., Portland, Oregon, USA). An intermittent
fault occurred in the probe installed at 5 cm depth at Site A; data from this probe are only available
for one of the CO2 sampling periods. Surface air temperature was measured 2 mm above the soil
surface (CS215-L, Campbell Scientific Inc., Logan, Utah, USA). All data were recorded every 2 minutes,
with 20 minute averages calculated and logged by a CR 1000 data logger (Campbell Scientific Inc.,
Logan, Utah, USA).
3.3.3 Soil CO2 sampling
Soil CO2 sampling was replicated in two adjacent polygons at each site. Within each polygon, five
opaque tubes (iPLex® Novadrain uPVC access tubes – 15 cm diameter by 16 cm high; iPLEX® DWV
136.150) were pressed into the soil surface to a depth of 6 cm. An opaque uPVC lid (iPLEX® DWV
136C.150) with an internal rubber O-ring, drilled to fit a pressure release bung and an injectable
septum (Baxter Interlink Injection Site Ref #2N3399K), was screwed on to the tubes immediately
before surface CO2 flux sampling. This created a chamber with a mean effective headspace of 0.0018
m3. In order to determine soil surface CO2 fluxes, 20 ml gas samples were taken from the headspace
chambers immediately after the chamber lids were fitted and vented, with three subsequent
37
samples taken at 20 min intervals over a 1 h period. Surface CO2 fluxes were calculated based on
headspace CO2 concentrations during the 1 h sampling period using the following equation:
𝑛𝑛𝐶𝐶𝑂𝑂2 = [𝐶𝐶𝑂𝑂2 ]
𝑃𝑃𝑃𝑃
𝑅𝑅𝑅𝑅𝑅𝑅
(3.1)
where nCO2 is the number of moles of CO2 emitted per m2 (µmol m-2), [CO2] is headspace CO2
concentration (µL L-1), P is atmospheric pressure (101325 Pa), V is the chamber volume (m3), R is the
ideal gas constant (8.314 J K-1 mol-1), T is air temperature (K), and A is the area covered by the
chamber (0.0177 m2). Surface CO2 flux rates (µmol m-2 s-1) were calculated by linear regression of
surface CO2 fluxes (µmol m-2) against time over which samples were collected (s). Linear regression
was also used to determine the rate of change of δ13CCO2 (‰ min-1) in the headspace chamber
samples.
Figure 3.5
Static headspace chamber and subsurface sampling tubes, Site B, Taylor Valley.
Soil CO2 sampling began > 96 h after installation of equipment. Gas samples were taken during the
warmest and coolest periods of the day, based on surface air temperature measurements at each
site, on two days in late December 2008. Site B was only sampled during one “cold” period due to an
38
error in determining the timing of the coldest part of the day. All CO2 samples were taken using gastight syringes (BD 30 ml syringe; Ref 309650) fitted with 3-way stopcocks (Baxter 2C6201) and 26G
½” needles (BD PrecisionGlide). Samples were injected into evacuated (0 atm) 12 ml vials
(Exetainers®, Labco Inc., UK) fitted with additional polytetrafluoroethylene (PTFE) septa (Supelco
Analytical, USA) to maintain sample integrity (Rochette and Bertrand, 2003; Knohl et al., 2004), and
stored in insulated containers prior to transport back to New Zealand. All samples were analysed for
CO2 concentration and δ13CCO2 using a trace gas module (TGII, Sercon Ltd., Cheshire, UK) connected to
a continuous flow isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK). The δ13C values are
expressed in parts per thousand (‰) enrichment or depletion of 13C, relative to Vienna Pee Dee
Belemnite (VPDB). Soil profile CO2 samples were analysed at Lincoln University, New Zealand (within
4 months of being collected), and surface CO2 flux samples at the University of California Davis Stable
Isotope Facility, California, USA (within 8 months of being collected). Standard reference gas samples
were also collected in the field following surface CO2 flux and subsurface CO2 sampling, and were
analysed with the samples. The standards showed no sign of storage effects. The analytical precision
of CO2 concentration and δ13C values was ≤5 ppm and ≤0.1‰, respectively.
3.3.4 Soil sampling, physical and chemical analyses
In polygons adjacent to CO2 sampling sites, soil pits were dug by hand to ice cement, and deepened
to a minimum of 55 cm depth using a concrete breaker to penetrate ice-cemented material. Soils
were described and classified using USDA nomenclature before being sampled volumetrically (500
cm3) at 5 cm contiguous intervals to 50 cm depth. All samples were returned to New Zealand where
they were air-dried, and weighed to determine bulk density (ρb) before being sieved (< 2 mm). All
chemical analyses were performed in duplicate on the < 2 mm fraction.
Soil pH and electrical conductivity were determined on a 1:5 soil: deionised water suspension (United
States Salinity Laboratory Staff, 1954). Samples were subsequently filtered through 0.45 µm syringe
filters (Advantec MFS-25) to remove suspended sediment prior to analysing for cations and anions.
Cations were determined by inductively coupled plasma optical emission spectroscopy (720-ES,
Varian, Melbourne, Australia) and anions by ion exchange chromatography (DX-120, Dionex,
Sunnyvale, California, USA). Total C and nitrogen (N) contents of solid samples were determined on a
finely ground (< 50 µm) subsample of the < 2 mm fraction. Samples were combusted at 1000°C using
a Europa ANCA-GSL CN analyser connected to a continuous flow isotope ratio mass spectrometer
which also determined δ13C and δ15N values (Sercon Ltd., Cheshire, UK). The δ13C and δ15N values are
expressed relative to VPDB and atmospheric N2, respectively. Two separate runs were performed in
order to obtain total soil δ13C and organic δ13C values, and to enable calculation of soil inorganic C
content. Samples analysed for total soil C, N, δ13C and δ15N were not subjected to any treatment.
39
Samples analysed for organic δ13C and total organic C content were fumigated with HCl to remove all
inorganic C prior to analysis. Subsamples of the finely ground fraction were put into small glass vials,
moistened with deionised water, and placed in a vacuum dessicator containing a beaker with 100 ml
of concentrated (12 M) HCl for 12 h (Harris et al., 2001). Samples were then dried at 60°C, cooled in a
desiccator and capped tightly prior to being weighed into tin capsules for analysis. Soil inorganic C
contents were calculated by subtracting the total C content of the HCl fumigated samples from the
total C values obtained from the untreated samples.
3.3.5 Statistical analyses
Principal component analysis (PCA) was used to identify the main components influencing the
variance in soil chemical properties with depth and between sites. PCA was completed using the
function princomp() in R 2.11.1 (The R Foundation). Linear regression was used in Microsoft Excel®
2007 to determine surface CO2 fluxes and the rate of change in δ13CCO2 in surface flux samples.
3.4 Results
3.4.1 Environmental variables
At the time of sampling (late December 2008), soils at Sites A and B were ice-cemented below 35 and
26 cm, respectively, although soil temperatures during the sampling periods were very similar at
both sites (Figure 3.6A and Figure 3.7A). The strongest vertical soil temperature gradients occurred
during “warm” sampling periods, with soil temperatures reaching 13−14°C at 5 cm depth, and
1.3−2.6°C at 25 cm depth. During “cold” sampling periods, soil temperature gradients were lower,
with temperatures ranging from ~0.5°C at 5 cm depth to ~1.6−3.6°C at 25 cm depth. The
progressively greater lag in soil temperature response to surface heating with increasing depth in the
profile meant that soil temperatures at 25 cm depth during the “cold” sampling period were slightly
warmer than those measured during the “warm” sampling period. The rate of change in soil
temperature during warm and cold sampling periods was always greatest at 5 cm depth, and
decreased with depth in the soil profile.
40
Figure 3.6
Soil temperature (A) and moisture content (B) during gas sampling periods, Site A Taylor
Valley. Note: data at 5 cm depth were only recorded during one sampling period due to
a faulty probe.
41
Figure 3.7
Soil temperature (A) and moisture content (B) during gas sampling periods, Site B Taylor
Valley. Note: this site was only sampled during one “cold” period due to an error in
determining the timing of the coldest part of the day.
Soil moisture content remained relatively constant throughout the sampling period at Site A (Figure
3.6B), with no moisture detected at 5 cm depth. Ponded water was observed on the soil surface in
low-lying areas within 300 m of the sampling area, suggesting subsurface melt from ice-cemented
soils on adjacent north-facing hill slopes was contributing to the relatively high moisture content
42
(~5%) observed at 15 cm depth at this site. Soil moisture content at 25 cm depth was around 6%. At
Site B, soil moisture content at 5 cm depth was higher during warm sampling periods, whereas at 15
cm depth it remained relatively constant (~1%) (Figure 3.7B). Soil moisture content at 25 cm depth
increased from 4 to 8% over the sampling period due to thawing ice-cement at the base of the
profile.
3.4.2 Soil physical and chemical properties
Soil bulk density ranged from 1.5 to 2.3 g cm-3. These values correspond to soil porosities ranging
from 18 to 45%, assuming a particle density of 2.8 g cm-3 (Barrett and Froggatt, 1978) which is
appropriate for the dominantly mafic igneous rocks of the study area.
Principal component analysis of soil chemical data revealed three principal components which
explained 80% of the variance in the original soil chemical data. Principal component (PC) 1
represents total C, inorganic C, calcium (Ca2+), magnesium (Mg2+), nitrate (NO3−) and sulphate (SO42−)
(Supplementary Table 3.4). PC 1 can be considered a C component as it is dominated by total C and
inorganic C. PC 2 represents pH, EC and sodium (Na+), and is effectively a measure of saltiness. PC 3
represents total N, organic C, potassium (K+), chloride (Cl−) and sulphate, although component
correlations suggest that total N and organic C are the variables best described by PC 3. Depth
profiles, contrasts and similarities between the soils at each site are discussed below in terms of the
major variables contributing to the three principal components.
Inorganic C (PC 1)
Inorganic C was higher at Site A than at Site B (Figure 3.8A). At Site A, inorganic C decreased from 2
mg g-1 to 1 mg g-1 in the upper 15 cm of the profile, and increased with depth below 15 cm. The
maximum inorganic C content (4 mg g-1) occurred in the 40−45 cm depth increment, coinciding with
the approximate maximum thaw depth of the active layer. Despite inorganic C content in the upper
15 cm being the lowest in the profile at Site A, it was more than double the organic C content. Below
15 cm depth, inorganic C was always more than four times greater than organic C at the same depth.
At Site B, inorganic C decreased from 2 mg g-1 in the top 5 cm to 0.3 mg g-1 between 10 and 25 cm
depth. Below 25 cm depth, inorganic C increased, and was always more than five times greater than
organic C at the same depth.
43
Figure 3.8
Depth profiles of A) soil inorganic C, B) soil organic C, and C) C isotopic composition of
soil organic C at Sites A and B, Taylor Valley. Error bars are ±1 standard deviation; n = 2.
Organic C (PC 3)
Total organic C decreased with depth down the profile at Site A (Figure 3.8B), except for a secondary
peak of 0.6 mg g-1 at 40−45 cm depth, which was only slightly lower than values measured between
0−10 cm depth. The depth of the secondary peak corresponds to the estimated maximum thaw
depth of the active layer. At Site B, total organic C values were very low (0.16 – 0.32 mg g-1) and
decreased slightly with depth.
From 5−15 cm depth at Site A, and in the upper 10 cm of the profile at Site B, δ13C values were highly
depleted relative to VPDB, ranging from −25.7 to −30.7‰ (Figure 3.8C). The upper 5 cm at Site A had
a comparatively enriched δ13C value of −21.6‰; below 15 cm depth, values ranged from −15.8 to
−19.6‰. At Site B, δ13C values beneath 10 cm depth were more depleted relative to those at Site A,
ranging from −18.6 to −20.6‰.
Soil pH and electrical conductivity (PC 2)
Soil pH ranged from ~ 10 in the upper part of the profile at both sites to between 8.8 and 9.4 at
45−50 cm depth (Figure 3.9A). Electrical conductivity (EC), a proxy for soil salt concentration, was
higher in the upper 10 cm of the soil at Site B than at Site A (Figure 3.9B). At both sites, EC values
were relatively low and constant between 10 and 40 cm depth. Between 40 and 50 cm depth, EC
values at both sites increased to a level similar to that in the upper 5 cm at Site A (Figure 3.9B).
44
Figure 3.9
Depth profiles of A) soil pH, and B) electrical conductivity at Sites A and B, Taylor Valley.
Error bars are ±1 standard deviation; n = 2.
3.4.3 CO2 dynamics
Subsurface CO2 profiles
Subsurface CO2 concentrations were significantly different between warm and cold periods at both
sites (Figure 3.10A). During warm periods, soil CO2 concentrations were greater than atmospheric
CO2 concentration (383.3 µL L-1 at the South Pole in December 2008; Keeling et al., 2001) in the upper
10 cm of the profile at Site A, reaching a maximum of 400 µL L-1 at 10 cm depth. Samples taken during
cold sampling periods showed soil CO2 concentrations were substantially below that of atmospheric
CO2, the lowest being 315 µL L-1 at 15 cm depth at Site A. At both sites, shifts in soil CO2 concentration
between warm and cold periods were mirrored about a trend-line that decreases from 380 µL L-1 at
the surface to 330−340 µL L-1 at 30 cm depth, demonstrating that cycles of CO2 uptake (cold periods)
and release (warm periods) occur. The amount of soil CO2 taken up and released was greater at Site
A than at Site B.
45
Figure 3.10 Soil profile CO2 concentration (A) and δ13CCO2 (B) at Sites A and B, Taylor Valley. Data
shown represent averages for warm and cold sampling periods (n = 6 for 0 cm samples
and n = 12 at all other depths, except for the cold sampling period at Site B, where n = 3
for 0 cm samples and n = 6 at all other depths). Error bars represent the SEM.
46
Soil profile δ13CCO2 was relatively enriched during warm periods when compared to cold periods at
both sites (Figure 3.10B). At Site A, δ13CCO2 became more enriched with increasing depth in the profile
during both warm and cold periods, although at 15 cm depth there were slight inflexions in the
trends. Values ranged from −7.0 to −9.0‰, with a maximum difference between warm and cold
periods of 1‰. At Site B, δ13CCO2 remained relatively constant with depth during both warm and cold
periods, ranging from −8.2 to −9.2‰, with differences of ≤ 0.9‰ between warm and cold periods.
Soil profile δ13CCO2 was similar to that of atmospheric CO2 (−8.22‰ at the South Pole in December
2008; Keeling et al., 2001) during warm periods at Site B. At both sites, δ13CCO2 was highly enriched
relative to soil organic matter (Figure 3.8C).
Surface CO2 fluxes
Surface CO2 flux rates ranged from −0.050 to 0.048 µmol m-2 s-1 (Table 3.1). Periods of efflux (positive
flux) and influx (negative flux) corresponded to warm and cold sampling periods, respectively. During
warm periods (efflux), headspace CO2 became progressively enriched (by up to 1.8‰), whereas
during cold periods (influx), headspace CO2 became depleted (by up to −2.0‰). The δ13CCO2 values
ranged from −6.5 to −10.6‰ during warm periods, and −7.4 to −11.2‰ during cold periods.
47
Surface soil CO2 fluxes and rate of change in δ13CCO2 during surface CO2 flux sampling at Sites A and B, Taylor Valley.
Table 3.1
a
Sampling
time a
CO2 flux
(µmol m-2 s-1)
Site
Chamber
A
a
W1
W2
C1
C2
0.010
0.000
-0.041**
-0.003
A
b
W1
W2
C1
C2
0.034
0.026
-0.024
-0.044
A
c
W1
W2
C1
C2
A
d
A
A
∆𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2
Sampling
time a
CO2 flux
(µmol m-2 s-1)
∆𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2
Site
Chamber
B
a
W1
W2
C1
0.025
0.034**
-0.037
0.022**
0.014
0.001
-0.002
B
b
W1
W2
C1
0.021
0.018**
-0.028*
0.021
0.011
-0.025
0.048
0.019
-0.039**
-0.038**
0.000
0.003*
-0.008
0.004
B
c
W1
W2
C1
0.045
no data
-0.050**
-0.002
no data
0.023
W1
W2
C1
C2
0.043**
0.005
0.004
-0.016
0.024**
0.016
-0.022**
-0.016*
B
d
W1
W2
C1
0.040*
0.010
-0.013
0.012
0.020*
-0.013
e
W1
W2
C1
C2
0.034*
0.024*
0.001
-0.039**
0.003
0.007
-0.051
-0.026**
B
e
W1
W2
C1
0.018
0.018**
-0.029*
-0.011
0.009
-0.008*
f
W1
W2
C1
C2
0.036
0.016**
-0.023**
-0.024*
0.008
0.029
-0.011**
-0.021*
B
f
W1
W2
C1
0.024*
0.013
0.001
0.004
0.011
-0.004
∆𝑡𝑡
0.010
0.013*
-0.006
-0.008
b
∆𝑡𝑡
0.015
0.010
-0.018
b
W and C denote warm and cold sampling periods, respectively. b Units are per mil min-1. * represents P <0.1; ** represents P <0.05.
48
3.5 Discussion
3.5.1 Soil CO2 uptake and release
The low soil CO2 flux rates measured in this study are in close agreement with in situ surface flux
rates in Taylor Valley reported by Burkins et al. (2001) and Parsons et al. (2004), but an order of
magnitude lower than those reported by Ball et al. (2009), despite all studies having used soils with
similar pH, EC, organic and inorganic C values. Higher flux rates such as those reported by Gregorich
et al. (2006) were from soils with organic C contents of up to 35 mg g-1. Such high organic C contents
are unusual in Antarctic Dry Valley soils; this study focuses on CO2 dynamics in typical Dry Valley soils
characterised by high pH (~10) and low organic C contents (<1 mg g-1).
It is well established that soil respiration increases with rising soil temperature (e.g. Lloyd and Taylor,
1994; Fang and Moncrieff, 2001). If CO2 release during warm periods at the study sites were due to
biological respiration, δ13CCO2 values should reflect the δ13C of the source organic material (Dörr and
Münnich, 1980). Soil δ13CCO2 values (Figure 3.10B) showed no relationship with soil organic matter
δ13C values (Figure 3.8C). Furthermore, if CO2 uptake during cold periods were a result of algal
photosynthesis, soil δ13CCO2 should be enriched due to preferential uptake of 12CO2 by photosynthetic
organisms, thus concentrating the heavier 13CO2 in the soil atmosphere (Park and Epstein, 1960). No
biological mechanism can be invoked to account for the results. Instead, δ13CCO2 values associated
with the uptake (depleted δ13CCO2) and release (enriched δ13CCO2) of soil CO2 indicate that CO2
dynamics in Dry Valley soils are controlled by other mechanisms.
In studies of surface CO2 fluxes in Taylor Valley, Ball et al. (2009) found that soil temperature
fluctuations (at 5 cm depth) accounted for almost half of the variation observed in diel CO2 fluxes. As
similar diel cycles of CO2 flux were also observed in heat-treated (120°C) soils, presumed to be free of
biological activity, Ball et al. (2009) suggested that dissolution of CO2 in soil water, consistent with
Henry’s Law, was a likely explanation for the diel variations observed. This hypothesis was tested by:
1) comparing soil profile CO2 concentrations measured at warm and cold periods with values
simulated by a model that incorporates Henry’s Law dissolution of CO2, and 2) comparing the
measured changes in δ13CCO2 with those expected from fractionation due to dissolution and
exsolution of CO2, biological respiration, and photosynthetic CO2 fixation. The data were also used to
quantify the relative contributions of biotic and abiotic processes to soil CO2 uptake and release.
A model of abiotic CO2 dynamics
At steady state, the closed-system partitioning of CO2 between gaseous and dissolved CO2 is
determined by Henry’s Law and solution pH. The temperature dependent Henry coefficient (KH)
(Stumm and Morgan, 1996) expresses the relationship between pCO2 of the atmosphere and
49
dissolved gaseous CO2 ([CO2(aq)] = [H2CO3]), as pCO2 = KH[CO2(aq)]. Total dissolved CO2 (CT), which
includes the species H2CO3, HCO3− and CO32−, is related to [CO2(aq)] and pH by:
𝐶𝐶𝑇𝑇 = �𝐶𝐶𝑂𝑂2(𝑎𝑎𝑎𝑎) � �1 +
𝐾𝐾𝑎𝑎1
𝐾𝐾𝑎𝑎 𝐾𝐾𝑎𝑎
+ 1+ 22 �
+
[𝐻𝐻 ] [𝐻𝐻 ]
(3.2)
where Ka1 and Ka2 are the equilibrium constants for [H+] [HCO3−]/[H2CO3] and [H+] [CO32−]/[HCO3−],
respectively (Stumm and Morgan, 1996). The temperature dependence of KH is such that as
temperature increases, [CO2(aq)] decreases. Thus, as a result of changes in KH, decreasing temperature
causes CO2 absorption, and increasing temperature causes CO2 desorption. In a soil, the capacity of
soil solution to absorb and desorb CO2 depends on soil solution volume, soil pH, and the magnitude
of soil temperature variation. From (3.2), it is evident that increasing soil solution pH (decreasing
[H+]) causes an increase in CT. A high soil solution pH interacts with the volume of soil solution,
amplifying the effect of CO2 adsorption or desorption with changing temperature by providing for a
larger source or sink of dissolved CO2 involved in the exchanges.
Using the principles described above, and closed-system, steady-state assumptions, abiotic CO2
dynamics in the soils at the study sites were modelled. The low flux rates measured (Table 3.1)
suggest the closed-system assumption is not unreasonable, but this is explored further below. The
model predicts the ratio of soil CO2 concentrations ([CO2]warm: [CO2]cold) when temperature (T) drops
from warm (Twarm) to cold (Tcold) in terms of the soil moisture content and pH of the system.
The ratio of subsurface soil CO2 concentrations attributable to abiotic processes as a result of a
change in temperature from Twarm to Tcold is given by:
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
�
+
�
[𝐶𝐶𝑂𝑂2 ]𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
𝑅𝑅𝑇𝑇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝐾𝐾𝐻𝐻𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝑇𝑇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
=
×
(𝜑𝜑 − 𝜃𝜃)
[𝐶𝐶𝑂𝑂2 ]𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝜃𝜃
𝑇𝑇𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
�𝑅𝑅𝑇𝑇
+ 𝐾𝐾
�
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
𝐻𝐻𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
(3.3)
where φ is soil porosity (L voids L soil-1), θ is soil volumetric moisture content (L water L soil-1), R is
the ideal gas constant (8.314 J K-1 mol-1), T is soil temperature (K), KH is Henry’s coefficient (L atm mol1
) and [CO2] is soil CO2 concentration (µL L-1). A full derivation of the model is given in Appendix A.
The dependence of [CO2]warm: [CO2]cold on soil pH and moisture content for temperature changes
(between warm and cold sampling periods) typical of those measured at 5 cm (Figure 3.11A) and 15
cm (Figure 3.11B) depth in the study area is shown in Figure 3.11. At low pH (≤ 6), CT is low (3.2), and
consequently a large soil solution volume (high moisture content) is required to provide the pool of
dissolved inorganic C necessary for temperature-driven absorption/desorption processes to shift
[CO2]warm: [CO2]cold significantly. Between pH 7 and 9, CT rises rapidly and the volume of soil solution
50
required to produce a given [CO2]warm: [CO2]cold declines rapidly. For example, at pH 6, soil moisture
content must be 0.32 (at or above field capacity in many soils) for [CO2]warm: [CO2]cold to be 1.4 (Twarm =
286 K and Tcold = 274 K; Figure 3.11A). By contrast, for the same change in T at pH 9, the same ratio
can be accommodated by a soil with a moisture content < 0.01 (< 1%). The magnitudes of [CO2]warm:
[CO2]cold simulated for 15 cm depth (Figure 3.11B) are much lower because of the smaller
temperature fluctuation at this depth, but the relative sensitivity to soil pH and moisture content
remains the same.
The model suggests that changes in soil profile CO2 concentrations in the study area should be
greatest at shallow soil depths, where temperature changes are greatest, pH is high (~10), and
moisture content is adequate. Conversely, greater depths should have smaller changes in soil CO2
concentrations due to smaller changes in soil temperature. However, the measured CO2
concentration depth profiles (Figure 3.10A) show relatively low variation in CO2 concentrations at
shallow depths, with maximum differences at 10−15 cm. As predicted by the model, little variation
occurs at the base of the profile.
51
Figure 3.11 Ratios of simulated [CO2]warm: [CO2]cold as influenced by soil moisture content and soil
solution pH. Ratios are calculated using equation 3.3. In (A) Twarm = 286 K and Tcold = 274
K which are similar to soil temperatures measured at 5 cm depth, and in (B) Twarm =
278.3 K and Tcold = 277.3 K, similar to soil temperatures measured at warm and cold
periods, respectively, at 15 cm depth in the study area. When the ratio of [CO2]warm:
[CO2]cold is 1, there is zero potential for CO2 to be absorbed. This occurs when soil
moisture is so low that no CO2 can be dissolved. Higher ratios represent greater changes
in soil CO2 concentrations between warm and cold periods, and thus greater uptake and
release of CO2 by soils.
52
Measured and simulated ratios of [CO2]warm: [CO2]cold were compared for 5, 15 and 25 cm depths
using measured values of soil pH, temperature and moisture content (Figure 3.12). The 25 cm depth
increment behaves as the model predicts. Ratios of [CO2]warm: [CO2]cold are very close to 1 due to
minimal changes in soil temperature between warm and cold periods. At this depth, the soil closely
approximates a closed system because ice-cemented soil beneath provides a gas-impermeable lower
barrier, and CO2 concentration gradients are low (Figure 3.10A). At 15 cm depth, measured ratios are
15% and 9% greater than simulated ratios (their theoretical abiotic maximum) at Sites A and B,
respectively, suggesting a possible biological respiration component. At 5 cm depth, measured
[CO2]warm: [CO2]cold ratios are considerably lower than simulated ratios. As the 5 cm depth increment
is most likely to suffer from violation of the closed-system assumption of the model, this discrepancy
is attributed to exchange between the soil and surface atmospheres. Leakage of CO2 from the soil
during warm periods when CO2 is released via exsolution, and influx of atmospheric CO2 during cold
periods when soil CO2 concentrations are depleted (CO2 dissolution) would act to damp variation in
measured CO2 concentrations. These inferred fluxes are supported by the soil CO2 concentration
gradients at this depth (Figure 3.10A).
Figure 3.12 Comparison between measured and simulated [CO2]warm: [CO2]cold ratios between warm
and cold sampling periods for Sites A and B, Taylor Valley.
53
Discriminating between biotic and abiotic processes using stable C isotopes
Biotic production and consumption of CO2 and the abiotic dissolution/exsolution processes outlined
above produce distinctive soil δ13CCO2 values. Biological respiration of highly depleted organic C
would produce respired CO2 of a very similar (highly depleted) C isotopic composition (Dörr and
Münnich, 1980), as respiration does not (or only to a minor extent) discriminate against 13C. Based on
measured soil organic δ13C values at both sites (Figure 3.8C), respired δ13CCO2 values of ~ −30 to
−21‰ in the upper 15 cm of the profiles, and ~ −20 to −16‰ below 15 cm depth would be expected.
Thus, if biological respiration were responsible for the increase in soil CO2 concentration during warm
periods, δ13CCO2 values would become relatively depleted. Conversely, discrimination against 13CO2 in
C3 photosynthetic pathways would produce a δ13CCO2-enriched atmosphere above actively
photosynthesising organisms. Therefore, if soil CO2 uptake during cold periods were due to
photosynthetic CO2 consumption, δ13CCO2 values would become relatively enriched. The results are
contrary to these patterns: soil profile δ13CCO2 values were relatively enriched during warm periods
(CO2 release) and depleted during cold periods (CO2 uptake; Figure 3.10B). Surface CO2 flux data also
showed progressive enrichment and depletion of δ13CCO2 during warm and cold sampling periods,
respectively (Table 3.1). The observed patterns are consistent with isotopic fractionation during CO2
dissolution and its speciation to HCO3−(aq) and CO32−(aq) (Table 3.2). At pH values >9, dissolved CO2
species are dominated by the HCO3− and CO32− forms (Stumm and Morgan, 1996). Equilibrium
fractionation of CO2 gas to HCO3− and CO32− results in a ~−10‰ depletion of CO2 gas relative to HCO3−
or CO32− (Thode et al., 1965; Mook et al., 1974; Table 2). Thus, dissolution of CO2 to form HCO3− and
CO32− (at the high pH values measured) would result in depletion of δ13CCO2 in the soil atmosphere,
whereas exsolution would lead to enrichment of δ13CCO2 in the soil atmosphere. The magnitudes of
the isotopic shifts observed (~ −1‰) are less than ~−10‰ because only a proportion of the CO2 in
the soil atmosphere exchanges with soil solution (~ 10%).
Table 3.2
Carbon isotope fractionation in the CO2−HCO3−CO3−CaCO3 system under equilibrium
conditions. 13εy/x represents the fractionation of compound y relative to compound x.
Soil temperatures (T) shown are similar to those measured during the sampling period. g
= gaseous CO2, a = dissolved CO2, b = dissolved HCO3−, c = dissolved CO32− ions, s = solid
calcium carbonate. TK = T (°C) + 273.15 K.
ε g/b b
(‰)
−10.70
13
1
εg/a a
(‰)
1.17
ε g/c c
(‰)
−10.06
ε g/s d
(‰)
−10.47
5
1.15
−10.20
−9.61
−10.19
T (°C)
13
13
13
a 13
εg/a = −0.19‰ + 373/TK (Vogel et al., 1970).
εg/b = 23.89‰ − 9483/TK (Mook et al., 1974).
c 13
εg/c = 13εg/b + 13εb/c (= −2.52‰ + 867/TK; Thode et al., 1965).
d 13
εg/s = 9.15‰ − 5380/TK (Mook, 2000).
b 13
54
Although the δ13CCO2 data confirm abiotic processes dominate CO2 dynamics in soils of Taylor Valley,
as postulated by Ball et al. (2009), the comparison of measured and simulated [CO2]warm: [CO2]cold
ratios suggests there is a component of CO2 flux around 15 cm depth that cannot be accounted for by
abiotic processes alone. Soil organic δ13C values in conjunction with soil profile δ13CCO2 values are
used to further constrain the potential biotic contribution to soil CO2 fluxes.
Quantifying the biotic contribution to soil CO2 fluxes
The bulge in soil CO2 concentration from 5−15 cm depth at Site A is likely to have a biological
respiration component, as organic C content at this depth is relatively high (~ 0.6 mg C g soil-1; Figure
3.8B) and the soil is relatively warm (Figure 3.6A). Assuming respired soil CO2 at this depth would
have an average δ13CCO2 value of −28.2‰ (Figure 3.8C), a mixing model (Fry, 2006) can be used to
calculate the biotic contribution (fbiotic) to soil CO2 fluxes:
𝑓𝑓𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 =
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 − 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
𝛿𝛿 13 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 − 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
(3.4)
where δ13CCO2-produced is the δ13CCO2 (‰) produced from cold to warm periods (Appendix B), δ13CCO2abiotic
is the δ13CCO2 (‰) of CO2 produced due to abiotic processes, and δ13Cbiotic is the δ13C (‰) of the
organic source material (Figure 3.8C).
This calculation requires the isotopic composition of abiotically produced CO2 (δ13CCO2-abiotic) to be
known. However, as it has not been empirically determined, fbiotic is calculated using a range of
δ13CCO2-abiotic values (Table 3.3) appropriate to different assumptions about isotopic fractionation
during dissolution and exsolution of CO2. The proportion of biotic CO2 production increases with
increasing values of δ13CCO2-abiotic. The most depleted value of δ13CCO2-abiotic (−8.22‰) assumes exsolved
CO2 has the same δ13CCO2 as atmospheric CO2 at the time of sampling. This is clearly unrealistic,
because firstly, it produces a negative fbiotic value, and secondly, exsolved CO2 this depleted would not
allow a shift of soil δ13CCO2 to > −8.22‰ as is measured at Site A (Figure 3.10B). A δ13CCO2-abiotic value of
−4.1‰ suggests a biotic contribution of 5%. This value of δ13CCO2-abiotic is determined using a mass
balance calculation of the C isotopic composition necessary to produce the shift in soil δ13CCO2 from
cold to warm sampling periods at Site B (see Appendix B for details of the calculation). Data from Site
B are used because the low organic C status of the soil is likely to limit respiration to the extent that
CO2 uptake and release throughout the whole profile is dominated by abiotic processes. However,
the model of abiotic soil CO2 uptake and release suggests there may be as much as a 9% biological
component to soil CO2 production at 15 cm depth at Site B. Therefore, conservatively assuming a
10% biological contribution from respiration of soil organic C to soil CO2 production over the whole
profile at Site B, the value of δ13CCO2-abiotic is −1.42‰ (Appendix B), which results in a biological
respiration component of 14%. The highest value of δ13CCO2-abiotic assumes no fractionation of CO2
55
released by exsolution from a 50:50 mix of HCO3− and CO32− in isotopic equilibrium with atmospheric
CO2. This is an unlikely scenario, but it establishes the most enriched possible value of δ13CCO2-abiotic.
The resultant biotic contribution to soil CO2 fluxes is 24%.
Table 3.3
Proportion of the change in CO2 concentration and isotopic composition attributable to
biotic processes (fbiotic) at Site A, 5−15 cm depth. This depth has the greatest shift in CO2
concentration between cold and warm periods (Figure 3.10A) and is the most likely
place in which biological respiration may occur. The effects of changes in assumed
δ13CCO2-abiotic values on fbiotic are shown.
fbiotic
δ13CCO2abiotic (‰)
−0.15
−8.22
0.05
−4.10
0.14
−1.42
0.24
2.16
Source of δ13CCO2-abiotic value
Atmospheric δ13CCO2 at the South Pole in December 2008 (Keeling
et al., 2001).
Average δ13CCO2 emitted from the whole profile at Site B,
assuming no biological contribution.
Assumes Site B has a 10% biological contribution, as suggested by
the comparison between measured and simulated [CO2]warm:
[CO2]cold ratios (Figure 3.12). The 10% biological contribution is
conservatively applied to the whole profile.
Assumes no fractionation of CO2 as it is exsolved from a 50:50
mix of HCO3−:CO32− in isotopic equilibrium with atmospheric CO2
(Table 3.2).
Implications for soil respiration measurements and C turnover times
Long-term records of soil respiration in the McMurdo Dry Valleys have been suggested as being one
of the most sensitive indicators of ecosystem response to climate variability (Barrett et al., 2006c).
However, it is essential that both biotic and abiotic contributions to soil CO2 fluxes are identified and
measured accurately, as both components are highly spatially and temporally variable. The
techniques utilised here represent a first attempt.
The data presented in this chapter demonstrate that previously measured in situ soil CO2 fluxes
attributed to biological respiration may be significant overestimates (e.g. Burkins et al., 2001; Barrett
et al., 2006b; Elberling et al., 2006; Gregorich et al., 2006), and C turnover times calculated on the
basis of such fluxes may be significant underestimates (e.g. Burkins et al., 2001; Elberling et al.,
2006). Based on their calculation of a mean organic C turnover time of 23 years, Burkins et al. (2001)
inferred that C cycling in Taylor Valley is non-steady-state, and suggested that at least two organic C
pools contribute to C cycling: a labile organic C pool replenished by in situ photosynthesis, and a
more recalcitrant organic C pool derived from ancient glacial and lake sediments. Such a short C
turnover time implies relatively low ecosystem resilience to fluctuations in C supply. The results from
this study, which show that biological respiration accounts for a (conservative) maximum of 25% of
measured soil CO2 fluxes in the Dry Valleys environment, mean that present estimates of C turnover
56
times could be at least four times greater, in the order of 100–500 years. The increased turnover
times resulting from lower rates of C utilisation imply that Dry Valley ecosystems are likely to be
more resilient to environmental change than what present estimates of C turnover time suggest.
3.6 Conclusions
Soil CO2 dynamics at the two sites studied in Taylor Valley are dominated by abiotic processes, with
small changes in soil temperature capable of driving significant changes in soil CO2 concentration.
The isotopic data imply that biological respiration makes only a minor contribution to soil CO2 fluxes
in the Dry Valleys environment.
These results have significant implications for understanding C cycling in polar desert environments.
The abiotic influence on soil CO2 dynamics must be accounted for and monitored, as CO2 effluxes
cannot be assumed to be solely biologically driven.
57
3.7 Supplementary information
3.7.1 Model of abiotic CO2 uptake and release
Model calculations were based on the following equations. Assuming a soil at steady state, at
temperature T (K) for an increment of depth D (m), the partial pressure of CO2, according to Henry’s
Law, is
𝑝𝑝𝑝𝑝𝑂𝑂2 = 𝐾𝐾𝐻𝐻𝑇𝑇 �𝐶𝐶𝑂𝑂2 (𝑎𝑎𝑎𝑎) � ,
(3.5)
where KH is the Henry coefficient (L atm mol-1) and [CO2(aq)] is dissolved gaseous CO2 (H2CO3).
Total dissolved CO2 (CT) is determined by the equilibrium constants for [H+] [HCO3−]/[H2CO3] and [H+]
[CO32−]/[HCO3−], Ka1 and Ka2, respectively.
𝐶𝐶𝑇𝑇 = �𝐶𝐶𝑂𝑂2(𝑎𝑎𝑎𝑎) � �1 +
𝐾𝐾
where 𝐾𝐾𝐻𝐻𝑇𝑇 ′ = 𝐾𝐾𝐻𝐻𝑇𝑇 ��1 + [𝐻𝐻𝑎𝑎1
+] +
𝐾𝐾𝑎𝑎1 𝐾𝐾𝑎𝑎2
[𝐻𝐻 + ]2
pressure of the soil atmosphere (atm).
𝐾𝐾𝑎𝑎1 𝐾𝐾𝑎𝑎1 𝐾𝐾𝑎𝑎2
𝑝𝑝𝑝𝑝𝑂𝑂2 𝑃𝑃[𝐶𝐶𝑂𝑂2 ]
+
�
=
=
,
[𝐻𝐻 + ] [𝐻𝐻 + ]2
𝐾𝐾𝐻𝐻𝑇𝑇 ′
𝐾𝐾𝐻𝐻𝑇𝑇 ′
(3.6)
� , [CO2] is the concentration of soil CO2 (µL L-1) and P is the
Per unit area of soil surface, the number of moles of dissolved CO2 (µmol) is given by
𝑃𝑃[𝐶𝐶𝑂𝑂2 ]
𝐷𝐷𝐷𝐷,
𝐾𝐾𝐻𝐻𝑇𝑇 ′
(3.7)
[𝐶𝐶𝑂𝑂2 ] 𝑃𝑃𝑉𝑉𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑃𝑃[𝐶𝐶𝑂𝑂2 ] 𝐷𝐷(𝜑𝜑 − 𝜃𝜃)
=
,
𝑅𝑅𝑅𝑅
𝑅𝑅𝑅𝑅
(3.8)
𝑛𝑛𝑛𝑛𝑂𝑂2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 =
where θ is the soil volumetric moisture content (L water L soil-1). The number of moles of CO2 in the
soil atmosphere (µmol)
𝑛𝑛𝑛𝑛𝑂𝑂2𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 =
where Vsoil atmos is the volume of the soil atmosphere (L), and φ is soil porosity (L voids L soil-1).
At temperature T1, total moles of CO2 in volume D
=
(𝜑𝜑 − 𝜃𝜃)
𝑃𝑃1 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1 𝐷𝐷(𝜑𝜑 − 𝜃𝜃) 𝑃𝑃1 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1
𝜃𝜃
+
𝐷𝐷𝐷𝐷 = 𝑃𝑃1 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1 𝐷𝐷 �
+
�.
𝑅𝑅𝑇𝑇1
𝐾𝐾𝐻𝐻𝑇𝑇1 ′
𝑅𝑅𝑇𝑇1
𝐾𝐾𝐻𝐻𝑇𝑇1 ′
(3.9)
58
Under the closed-system assumption, conservation of mass dictates that between T1 and T2,
and therefore
𝑃𝑃2 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇2 𝐷𝐷 �
(𝜑𝜑 − 𝜃𝜃)
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
𝜃𝜃
+
+
� = 𝑃𝑃1 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1 𝐷𝐷 �
�,
𝑅𝑅𝑇𝑇2
𝐾𝐾𝐻𝐻𝑇𝑇2 ′
𝑅𝑅𝑇𝑇1
𝐾𝐾𝐻𝐻𝑇𝑇1 ′
(3.10)
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
� 𝑅𝑅𝑇𝑇 +
�
𝐾𝐾𝐻𝐻𝑇𝑇2 ′
2
(3.11)
𝑃𝑃1 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1
=
.
𝑃𝑃2 [𝐶𝐶𝑂𝑂2 ] 𝑇𝑇2
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
� 𝑅𝑅𝑇𝑇 +
�
𝐾𝐾𝐻𝐻𝑇𝑇1 ′
1
Given the exchanges of CO2 between atmosphere and solution phases are small in relation to the
total mass of the soil atmosphere, from the ideal gas law, the ratio of the pressures of the soil
atmosphere between T1 and T2 is equal to the ratio of the temperatures, and therefore
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
� 𝑅𝑅𝑇𝑇 +
�
𝐾𝐾𝐻𝐻𝑇𝑇2 ′
2
(3.12)
[𝐶𝐶𝑂𝑂2 ] 𝑇𝑇1
𝑇𝑇2
=
× .
[𝐶𝐶𝑂𝑂2 ] 𝑇𝑇2
𝑇𝑇1
(𝜑𝜑 − 𝜃𝜃)
𝜃𝜃
� 𝑅𝑅𝑇𝑇 +
�
𝐾𝐾𝐻𝐻𝑇𝑇1 ′
1
3.7.2 Calculating the isotopic composition of abiotically-produced CO2
(δ13CCO2-abiotic)
Using a mass balance approach, the average δ13CCO2 of the CO2 produced between cold and warm
sampling periods must satisfy the following
𝐷𝐷
� 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 [𝐶𝐶𝑂𝑂2 ]𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑑𝑑𝑑𝑑 + ����������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 [𝐶𝐶𝑂𝑂2 ]𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
0
(3.13)
𝐷𝐷
= � 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 [𝐶𝐶𝑂𝑂2 ]𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 𝑑𝑑𝑑𝑑,
0
where z is soil depth (m), D is the depth to ice cement (m), [CO2] is soil CO2 concentration (µL L-1) and
����������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 is the concentration-weighted mean δ13CCO2 of the produced CO2 (‰).
The term [𝐶𝐶𝑂𝑂2 ]𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 was calculated as
𝐷𝐷
𝐷𝐷
� [𝐶𝐶𝑂𝑂2 ]𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 𝑑𝑑𝑑𝑑 − � [𝐶𝐶𝑂𝑂2 ]𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑑𝑑𝑑𝑑.
0
(3.14)
0
Assuming there is no biological contribution to the CO2 produced at Site B,
��������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = ����������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 .
(3.15)
59
If, however, there is a 10% biological contribution to the CO2 produced at Site B, and assuming that
13 𝐶𝐶
��������������������
δ13CCO2 reflects the δ13C of the organic source material (δ13Cbiotic), then 𝛿𝛿
𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 (the mean
δ13CCO2 (‰) of CO2 produced due to abiotic processes) can be derived as
����������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 − 0.1 (𝛿𝛿 13 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 )
��������������������
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 =
.
0.9
(3.16)
Values used to calculate δ13CCO2-abiotic are [CO2]produced = 29 µL L-1, δ13CCO2-produced = −4.1‰ and δ13Cbiotic =
−28.2‰. The integrals were approximated from the point data shown in Figure 3.10.
3.7.3 Results of principal components analysis
Table 3.4
Eigenvector coefficients and component correlations for the first three principal
components. Highlighted cells represent values higher than Mardia’s criterion (0.7 times
the highest coefficient for the component) for the eigenvector coefficients. Highlighted
cells in the component correlations columns show variables that are well described by
the components.
Variable
pH
EC
Total C
Organic C
δ13C organic
Inorganic C
δ13C inorganic
Total N
δ15N
CN ratio
Ca2+
K+
Mg2+
Na+
ClNO3PO42SO42-
Eigenvector coefficients
Comp.1
Comp.2
Comp.3
0.172
-0.369
0.128
-0.193
-0.325
-0.249
-0.369
0.092
0.144
-0.252
-0.075
0.374
-0.031
0.376
-0.089
-0.365
0.117
0.09
-0.203
-0.253
0.124
-0.207
-0.104
0.401
0.086
0.219
-0.198
0.231
0.124
-0.345
-0.29
0.229
-0.228
-0.136
-0.245
-0.347
-0.342
0.144
-0.195
-0.079
-0.401
-0.157
-0.185
-0.274
-0.301
-0.322
0.06
0.037
0.152
-0.278
0.007
-0.261
-0.044
-0.289
Component correlations
Comp.1
Comp.2
Comp.3
0.418
-0.84
0.234
-0.469
-0.742
-0.456
-0.898
0.209
0.263
-0.613
-0.171
0.685
-0.076
0.857
-0.163
-0.888
0.266
0.166
-0.493
-0.578
0.227
-0.504
-0.236
0.733
0.209
0.499
-0.362
0.561
0.283
-0.631
-0.705
0.522
-0.418
-0.33
-0.558
-0.634
-0.831
0.328
-0.357
-0.191
-0.914
-0.287
-0.449
-0.625
-0.551
-0.782
0.138
0.068
0.369
-0.634
0.013
-0.634
-0.1
-0.529
60
Chapter 4
Quantifying diel variations in biotic and abiotic soil CO2 fluxes,
Taylor Valley, Antarctica (I): Background rates in the absence of a
contemporary organic carbon source
4.1 Introduction
In Chapter 3, assumptions of a closed system (no atmospheric exchange) and steady-state
partitioning between gaseous and dissolved inorganic C according to Henry’s Law were used to
model abiotic subsoil CO2 dynamics. Comparisons with data measured at the warmest and coldest
parts of the day suggested this mechanism could account for subsoil CO2 dynamics. Under the closed
system assumption, changes in the C isotopic composition of soil CO2 between the cold and warm
time of the day in a relatively high organic C soil (Site A) limited a biological contribution to soil CO2
fluxes to less than 25%, depending on the assumptions made about the isotopic composition of CO2
exsolved from soil solution.
This chapter aims to further advance the understanding of Dry Valley soil CO2 dynamics through the
analysis of samples taken during a more intensive field sampling regime, in which samples were
taken at four-hourly intervals over a 48-h period. Sampling was conducted at Site B in Taylor Valley,
which is the same “Site B” as that studied in Chapter 3. The higher temporal resolution of the data,
combined with the coupling of soil CO2 concentration and δ13CCO2 profiles with surface CO2 fluxes,
provides the first comprehensive study of CO2 dynamics in Antarctic soils.
The objectives of the chapter are to:
1. Corroborate existing information on diel variation in surface CO2 fluxes;
2. Quantify surface and subsurface (storage) fluxes in order to test the closed system
assumption used in Chapter 3;
3. Partition biotic and abiotic contributions to surface soil CO2 fluxes by exploiting differences in
their C isotopic composition.
The field work which generated the data considered in this chapter was undertaken during the
2009/10 austral summer, a full year after the field work which yielded results for the previous
chapter. The materials and methods utilised were essentially the same as those employed during the
2008/09 summer, although specific differences are noted below.
61
4.2 Materials and methods
4.2.1 Study area
At the time of gas sampling, soils at Site B were ice-cemented below 30 cm depth, as opposed to
being ice-cemented below 26 cm during the sampling period in the 2008/09 austral summer.
4.2.2 Environmental variables
During the intensive 48-h gas sampling period, all data were recorded and logged every 2 minutes by
a CR850 data logger (Campbell Scientific Inc., Logan, Utah, USA).
Soil temperature and moisture content measurements
Soil temperature and volumetric moisture content at 5, 15 and 25 cm depth were measured using
Hydra Probes (Hydra Probe II, Stevens Water Monitoring Systems Inc., Portland, Oregon, USA).
The rate of change in soil temperature (ΔT/Δtt) was calculated as the difference in temperatures (T)
measured ten minutes apart, as follows:
Air temperature
∆𝑇𝑇 𝑇𝑇𝑡𝑡+5 𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑇𝑇𝑡𝑡−5 𝑚𝑚𝑚𝑚𝑚𝑚
=
.
∆𝑡𝑡𝑡𝑡
10 𝑚𝑚𝑚𝑚𝑚𝑚
(4.1)
Surface air temperature was measured 2 mm above the soil surface (CS215-L, Campbell Scientific
Inc., Logan, Utah, USA).
Wind speed
Wind speed (#40C Anemometer, NRG Systems, Hinesburg, Vermont, USA) and direction (W200P
Potentiometer Windvane, Vector Instruments, North Wales, UK) were measured 1.2 m and 1.3 m
above the soil surface, respectively. These measurements were made at Site A, located 2.8 km east
of Site B. As Site B was on a moraine crest, it is almost certain that wind speeds at Site B would have
been the same as or greater than those measured at Site A.
4.2.3 Soil CO2 sampling and analysis
The experimental set-up and sampling for surface soil CO2 fluxes proceeded as in 2008/09 (Chapter
3), except that only three chambers were installed in each of two adjacent polygons at Site B. The
reduced number of chambers enabled four 20 ml gas samples to be taken from each chamber at 15
min intervals over a 45 min period. The polygons in which gas samples were taken were located ~ 80
m west of the sampling site utilised in 2008/09, on the same geomorphic surface.
62
Soil profile CO2 sampling also proceeded as per the previous summer, with subsurface sampling
tubes installed adjacent to each of the three chambers in each polygon at depths of 5, 10, 15, 20 and
30 cm. Additionally, one sampling tube was installed horizontally, 1 cm above the desert pavement,
in each polygon. Samples were taken from these tubes in the same manner as the subsurface
sampling tubes, and provided an atmospheric CO2 sample from just above the soil surface.
Soil CO2 sampling began > 48 h after installation of equipment. Gas samples were taken every four
hours over a 48-h period, starting at 14:30 h on December 29th, 2009. A selection of samples were
analysed for CO2 concentration and δ13CCO2 using a Gas-bench II connected to a DeltaPlus Advantage
isotope ratio mass spectrometer (both Thermo Finnigan, Bremen, Germany) at the Macaulay Land
Use Research Institute in Aberdeen, Scotland, within 5 months of being collected. Budget constraints
meant that it was not possible to analyse all of the samples collected. Standard reference gas
samples collected in the field and standard reference gases prepared in New Zealand were included
with samples sent to Scotland, and showed no sign of storage or transport effects. The analytical
precision for CO2 concentration and δ13CCO2 values was ≤ 5 µL L-1 and ≤ 0.05‰, respectively.
4.2.4 Soil sampling, physical and chemical analyses
The gas samples at Site B were taken within 75 m of the soil pit sampled at Site B in 2008/09.
Additional soil samples were taken from a soil pit located adjacent to the 2009/10 gas sampling site,
although budget constraints meant that these samples were not analysed. However, based on the
proximity of the 2009/10 gas sampling site to the previously sampled and analysed soil pit, the
uniformity of the geomorphic surface on which samples were taken, and the similar physical
appearance of soil pits dug and described in polygons adjacent to the 2009/10 gas sampling site
(Figure 4.1), it was considered that the soil physical and chemical characteristics at Site B are
accurately represented by the data presented in Chapter 3.
63
Figure 4.1
Morphological comparison between soil pits adjacent to gas sampling sites at Site B in
the 2008/09 and 2009/10 austral summers. At the time of sampling, the soils (Typic
Haploturbels) were ice-cemented below 26 and 30 cm depth in 2008/09 and 2009/10,
respectively. Both profiles have been deepened with a concrete breaker and are
moistened from melting of ice-cemented permafrost.
4.2.5 Statistical analyses and calculations
Surface CO2 fluxes
Surface CO2 fluxes were determined from changes in headspace CO2 concentrations during the 45
min sampling period using the following equation:
𝑛𝑛𝐶𝐶𝐶𝐶2 = [𝐶𝐶𝑂𝑂2 ]
𝑃𝑃𝑃𝑃
𝑅𝑅𝑅𝑅𝑅𝑅
(4.2)
where nCO2 is the number of moles of CO2 emitted per m2 (µmol m-2), [CO2] is headspace CO2
concentration (µL L-1), P is atmospheric pressure (101325 Pa), V is the chamber volume (m3), R is the
ideal gas constant (8.314 J K-1 mol-1), T is air temperature (K), and A is the area covered by the
chamber (0.0177 m2).
Surface CO2 flux rates (µmol m-2 s-1) were calculated by linear regression (Microsoft Excel® 2010) of
surface CO2 fluxes (µmol m-2) against time over which samples were collected (s). Surface flux rates
that were significantly different to zero (p < 0.1) were subsequently used to calculate the net CO2
gained or lost from the surface chambers (Δ[CO2]), and the average δ13C of the net CO2 gained or lost
(δ13CΔCO2). Each significant surface CO2 flux, which may be positive or negative, was considered
individually, and calculations were as follows:
64
∆[𝐶𝐶𝑂𝑂2 ] = [𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖
(4.3)
where [CO2]final and [CO2]initial are the final and initial CO2 concentrations (µL L-1) measured from the
final and initial CO2 samples taken from the chamber, respectively.
Using a mass balance approach,
∆[𝐶𝐶𝑂𝑂2 ] ∙ 𝛿𝛿 13 𝐶𝐶∆𝐶𝐶𝐶𝐶2 = [𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ,
(4.4)
where δ13CΔCO2 (‰) is the δ13C of the net CO2 gained or lost between the final and initial
measurements, and δ13C[CO2]final and δ13C[CO2]initial are the δ13CCO2 values (‰) measured from the final
and initial CO2 samples taken from the chamber, respectively.
Therefore,
𝛿𝛿 13 𝐶𝐶∆𝐶𝐶𝐶𝐶2 =
[𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖
∙
∆[𝐶𝐶𝑂𝑂2 ]
(4.5)
Subsurface CO2 fluxes
Subsurface CO2 fluxes were estimated by calculating differences in the average soil profile CO2
concentration between consecutive sampling times. The average soil profile CO2 concentration was
determined by using an analogue integration technique whereby the area under a paper plot of CO2
concentration versus depth at each sampling time was weighed to determine the integral
30
∫ [𝐶𝐶𝑂𝑂2 ] 𝑑𝑑𝑑𝑑
��������
[𝐶𝐶𝑂𝑂2 ] = 0
30
(4.6)
��������
where z is depth in cm and [𝐶𝐶𝑂𝑂
2 ] is the profile average CO2 concentration between 0 and 30 cm
��������
depth. The difference in [𝐶𝐶𝑂𝑂
2 ] between consecutive sampling times was used to determine the
number of moles of CO2 (nCO2; µmol m-2) gained or lost over the period between sampling times using
the ideal gas law:
𝑛𝑛𝐶𝐶𝐶𝐶2 = ∆[𝐶𝐶𝑂𝑂2 ]
𝑃𝑃𝑃𝑃
𝑅𝑅𝑅𝑅
(4.7)
where Δ[CO2] is the difference in the profile average CO2 concentration between sampling times (µL
L-1), P is atmospheric pressure (101325 Pa), V is the soil gas volume (m3 m-2; calculated by multiplying
soil depth (0.3 m) by the air-filled porosity (ε = 0.4)), R is the ideal gas constant (8.314 J K-1 mol-1) and
T is soil temperature (K). This was converted to a flux rate (µmol m-2 s-1) by dividing by the time (s)
over which nCO2 was gained or lost.
65
Calculated surface CO2 fluxes and their C isotopic composition
To test that gas fluxes were dominantly diffusion-driven, surface CO2 flux rates were plotted against
near-surface CO2 concentration gradients. Linear regression of this relationship yields a value for the
effective diffusivity of CO2 (De; cm2 s-1) within the soil.
In a diffusion-driven system, the δ13CΔCO2 of surface CO2 fluxes is determined by subsurface CO2
concentration and δ13CCO2 gradients. Using data from subsurface profiles, the δ13CΔCO2 of surface CO2
fluxes can therefore be determined as follows:
𝑅𝑅13 𝑜𝑜𝑜𝑜 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 =
𝐹𝐹13
𝐹𝐹12
(4.8)
where R13 of flux is the ratio of 13C to 12C in the CO2 flux, and F13 and F12 refer to the flux of 13CO2 and
12
CO2, respectively. Each individual flux of 13CO2 and 12CO2 is derived separately, using Fick’s Law, so:
𝐹𝐹13 =
13
𝐷𝐷𝑠𝑠
𝑑𝑑13 𝐶𝐶𝑂𝑂2
, 𝑎𝑎𝑎𝑎𝑎𝑎 𝐹𝐹12 =
𝑑𝑑𝑑𝑑
12
𝐷𝐷𝑠𝑠
𝑑𝑑12 𝐶𝐶𝑂𝑂2
𝑑𝑑𝑑𝑑
(4.9)
where 13Ds and 12Ds are the diffusivities of 13CO2 and 12CO2 in soil, respectively, and d13CO2/dz and
d12CO2/dz are concentration gradients of 13C and 12C, respectively, where depth (z) increases
downwards in the soil. As such, the negative sign normally seen in Fick’s Law is not appropriate.
The ratio of 13C to 12C in the flux can therefore be derived as follows:
𝐹𝐹13
𝐹𝐹12
𝑑𝑑13 𝐶𝐶𝑂𝑂2
𝐷𝐷𝑠𝑠
𝑑𝑑𝑑𝑑 .
= 12
𝐷𝐷𝑠𝑠 𝑑𝑑12 𝐶𝐶𝑂𝑂2
𝑑𝑑𝑑𝑑
13
(4.10)
Any value presented in delta notation (e.g. δ13CCO2, in units of per mil) is related to a standard
reference material as follows:
𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝛿𝛿𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = �
� 1000 ,
𝑅𝑅𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
(4.11)
where Rsample and Rstandard are the ratios of heavy to light isotopes in the sample and standard,
respectively. Therefore, a δ13CCO2 value can be presented as:
𝐽𝐽
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2 = �
− 1� 1000 ,
𝑅𝑅𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
(4.12)
where J is the ratio of 13C to 12C in the sample.
66
Therefore,
𝐽𝐽 = �
𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2
+ 1� 𝑅𝑅𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 .
1000
(4.13)
The concentration of 13CO2 ([13CO2]) can be determined by
� 13𝐶𝐶 𝑂𝑂2 � =
and the concentration of 12CO2 is
𝐽𝐽
[𝐶𝐶𝑂𝑂2 ] ,
1 + 𝐽𝐽
(4.14)
[𝐶𝐶𝑂𝑂2 ]
.
1 + 𝐽𝐽
(4.15)
� 12𝐶𝐶 𝑂𝑂2 � =
Therefore, the flux of 13CO2 is
𝑑𝑑� 13𝐶𝐶 𝑂𝑂2 �
𝐷𝐷𝑠𝑠
=
𝑑𝑑𝑑𝑑
13
and the flux of 12CO2 is
𝑑𝑑� 12𝐶𝐶 𝑂𝑂2 �
𝐷𝐷𝑠𝑠
=
𝑑𝑑𝑑𝑑
12
𝐽𝐽
[𝐶𝐶𝑂𝑂2 ]�
𝑑𝑑 �
1 + 𝐽𝐽
𝐷𝐷𝑠𝑠
,
𝑑𝑑𝑑𝑑
(4.16)
13
12
𝐷𝐷𝑠𝑠
[𝐶𝐶𝑂𝑂 ]
𝑑𝑑 � 1 +2𝐽𝐽 �
𝑑𝑑𝑑𝑑
(4.17)
.
Using the derivative of a product rule,
𝐽𝐽
𝑑𝑑 �1 + 𝐽𝐽�
𝑑𝑑� 13𝐶𝐶 𝑂𝑂2 �
𝐽𝐽 𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
= [𝐶𝐶𝑂𝑂2 ]
+
.
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
1 + 𝐽𝐽 𝑑𝑑𝑑𝑑
(4.18)
[𝐶𝐶𝑂𝑂2 ] 𝑑𝑑𝐽𝐽
𝑑𝑑� 13𝐶𝐶 𝑂𝑂2 �
𝐽𝐽 𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
=
+
.
2
(1 + 𝐽𝐽) 𝑑𝑑𝑑𝑑 1 + 𝐽𝐽 𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(4.19)
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝑑𝑑
𝑑𝑑� 12𝐶𝐶 𝑂𝑂2 � (1 + 𝐽𝐽) 𝑑𝑑𝑑𝑑 − [𝐶𝐶𝑂𝑂2 ] 𝑑𝑑𝑑𝑑
=
,
(1 + 𝐽𝐽)2
𝑑𝑑𝑑𝑑
(4.20)
Using the rule for differentiation of a quotient for the first term on the right hand side of equation
(4.18) (above),
Again, using the quotient rule for differentiation of equation (4.17),
67
hence
𝑑𝑑� 13𝐶𝐶 𝑂𝑂2 � 12 𝑑𝑑� 12𝐶𝐶 𝑂𝑂2 �
𝐷𝐷𝑠𝑠
� 𝐷𝐷𝑠𝑠
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(4.21)
13
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝑑𝑑
(1 + 𝐽𝐽)
− [𝐶𝐶𝑂𝑂2 ]
[𝐶𝐶𝑂𝑂2 ] 𝑑𝑑𝐽𝐽
𝐽𝐽 𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
= 12 �
+
�
���
(1 + 𝐽𝐽)2
𝐷𝐷𝑠𝑠 (1 + 𝐽𝐽)2 𝑑𝑑𝑑𝑑 1 + 𝐽𝐽 𝑑𝑑𝑑𝑑
13
𝐷𝐷𝑠𝑠
=
=
Therefore,
𝑑𝑑𝐽𝐽
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝑑𝑑
[𝐶𝐶𝑂𝑂2 ] + 𝐽𝐽(1 + 𝐽𝐽)
− [𝐶𝐶𝑂𝑂2 ] �
���(1 + 𝐽𝐽)
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(4.22)
𝐷𝐷𝑠𝑠 𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝐽𝐽
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝐽𝐽
(1 + 𝐽𝐽) − [𝐶𝐶𝑂𝑂2 ]� .
𝐽𝐽(1 + 𝐽𝐽) + [𝐶𝐶𝑂𝑂2 ]�� �
�
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
𝐷𝐷𝑠𝑠
(4.23)
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑑𝑑𝐽𝐽
𝐷𝐷𝑠𝑠 𝑑𝑑𝑑𝑑 𝐽𝐽(1 + 𝐽𝐽) + 𝑑𝑑𝑑𝑑 [𝐶𝐶𝑂𝑂2 ]
= 12
.
𝐷𝐷𝑠𝑠 𝑑𝑑[𝐶𝐶𝑂𝑂2 ] (1 + 𝐽𝐽) − 𝑑𝑑𝐽𝐽 [𝐶𝐶𝑂𝑂 ]
2
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(4.24)
13
𝐷𝐷𝑠𝑠
12
𝐷𝐷𝑠𝑠
�
13
12
𝑅𝑅13
13
13
𝐷𝐷
𝐷𝐷
13
Assuming 12𝐷𝐷𝑠𝑠 scales as 12𝐷𝐷, the ratio is given by
𝑠𝑠
1
𝐷𝐷( 13𝐶𝐶𝐶𝐶2 )
𝑀𝑀( 12𝐶𝐶𝐶𝐶2 ) 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 𝑀𝑀( 13𝐶𝐶𝐶𝐶2 ) + 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 2
=
�
� = 0.9956.
𝐷𝐷( 12𝐶𝐶𝐶𝐶2 )
𝑀𝑀( 12𝐶𝐶𝐶𝐶2 ) + 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 𝑀𝑀( 13𝐶𝐶𝐶𝐶2 ) 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎
(Craig, 1953)
(4.25)
Therefore,
𝑅𝑅13
𝑑𝑑(𝛿𝛿 13 𝐶𝐶)
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑅𝑅
[𝐶𝐶𝑂𝑂2 ]
𝐽𝐽(1 + 𝐽𝐽) + 𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
1000
𝑑𝑑𝑑𝑑
= 0.9956 � 𝑑𝑑𝑑𝑑
�.
13
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑅𝑅𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑑𝑑(𝛿𝛿 𝐶𝐶)
[𝐶𝐶𝑂𝑂2 ]
(1 + 𝐽𝐽) −
1000
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(4.26)
R13 can be converted to standard delta notation using equation 4.11. R13 is substituted for Rsample, and
RVPDB is Rstandard.
68
4.3 Results
4.3.1 Environmental variables
At the time of sampling (late December 2009), soils at Site B were ice-cemented below 30 cm. Air
and soil temperatures varied diurnally, with the latter exhibiting progressively greater lags and
damping in soil temperature response to surface heating and cooling with increasing depth in the
profile (Figure 4.2A). Air temperature ranged from −4.4 to 6.1°C, and soil temperatures at 5 cm, 15
cm and 25 cm ranged from −1.5 to 11.7°C, 0.0 to 3.8°C and −0.4 to 0.8°C, respectively.
Soil moisture content was very low (Figure 4.2B), averaging 0.6, 0.4 and 1.3% at 5, 15 and 25 cm
depth, respectively. Qualitative observations of limited melting of snow patches and lake ice in late
December 2009 suggested that the summer was cooler than the previous summer of 2008/09. This
was corroborated by the lack of ponded surface water evident at sites where it had been observed
the previous summer, and the low soil moisture contents measured at Site B.
69
Figure 4.2
Soil temperature (A) and volumetric moisture content (B) during the 48-hour period
over which gas samples were taken, Site B, Taylor Valley. Data plotted for moisture
content are 30 min. moving averages. The narrow vertical grey boxes denote the periods
during which gas samples were taken.
70
4.3.2 CO2 dynamics
Surface CO2 fluxes
Surface CO2 flux rates at Site B were relatively low, with statistically significant fluxes (which
comprised 81% of all surface CO2 fluxes calculated) ranging from −0.053 to 0.074 µmol m-2 s-1 (Table
4.1). The average significant flux at each sampling time ranged between −0.041 and 0.058 µmol m-2
s-1 (Figure 4.3A). The highest and lowest (most negative) surface CO2 flux rates occurred at 14:30 and
02:30 h, respectively. Temporal patterns of variation in surface CO2 flux rates were similar to diel
trends in wind speed (measured at Site A) and soil temperature at 5 cm depth, although changes in
surface CO2 flux rate led variations in soil temperature at 15 cm depth by around 4 h (Figure 4.3C).
However, changes in surface CO2 flux rates were in phase with variations in the rate of change of soil
temperature at 15 cm depth (Figure 4.3D). The diel variation in significant surface CO2 flux rates was
asymmetrical about 0 µmol m-2 s-1, with maximum positive fluxes, measured at the warmest time of
day, being greater than minimum (negative) fluxes measured at the coolest time of day (Figure 4.3A).
71
Figure 4.3
(A) Average significant surface and subsurface soil CO2 flux rates at Site B, Taylor Valley.
Error bars represent the standard error of the mean. Subsurface soil CO2 storage fluxes
are shown as grey bars which represent the net CO2 flux over the 4-h period between
72
sampling. (B) Isotopic composition (δ13CΔCO2) of measured and predicted surface CO2
fluxes at Site B, Taylor Valley. Predicted values are calculated based on subsurface CO2
concentration and δ13CCO2 gradients at the time of sampling, according to equation 4.26.
Error bars represent the standard error of the mean. The dashed horizontal line
represents the average δ13CCO2 of soil CO2 (−8.8‰). (C) Soil temperature at 5 and 15 cm
depth (left hand axis) during the gas sampling period, Site B, Taylor Valley. Wind speed
data (right hand axis) are from Site A, Taylor Valley. (D) Rate of change in soil
temperature at 5 and 15 cm depth during the gas sampling period, Site B, Taylor Valley.
Data plotted are 30 min moving averages.
The dashed vertical lines correspond to the four-hourly intervals at which gas sampling
began.
Table 4.1
Date
29/12/10
Surface soil CO2 fluxes at Site B, Taylor Valley.
Day 1
Sampling Replicate
time (h)
14:30
a
b
c
d
CO2 flux
Date
0.063**
0.028
0.006
0.053**
30/12/10
(µmol m-2 s-1)
Day 2
Sampling Replicate
time (h)
14:30
a
b
c
d
CO2 flux
(µmol m-2 s-1)
0.046**
0.043**
0.067**
0.070**
29/12/10
18:30
a
b
c
d
0.056**
0.033**
0.074**
0.056
30/12/10
18:30
a
b
c
d
−0.008
−0.010**
−0.009*
0.020**
29/12/10
22:30
a
b
c
d
−0.017*
−0.045**
−0.015**
−0.026**
30/12/10
22:30
a
b
c
d
−0.015**
−0.028**
−0.024**
−0.032**
30/12/10
02:30
a
b
c
d
−0.037**
−0.032**
−0.053**
−0.034**
31/12/10
02:30
a
b
c
d
−0.051**
−0.039*
−0.029*
−0.045**
30/12/10
06:30
a
b
c
d
−0.010
−0.037*
−0.025*
−0.024**
31/12/10
06:30
a
b
c
d
−0.010
0.015**
0.039**
0.002
30/12/10
10:30
a
b
c
d
0.023*
0.028**
0.034**
0.030**
31/12/10
10:30
a
b
c
d
0.039**
0.038
0.019
0.050**
* represents P < 0.1; ** represents P < 0.05.
73
The average δ13CCO2 of the four gas samples taken from surface chambers during the 45-min sampling
periods varied little from that of ambient atmospheric CO2 (−8.8‰), ranging between −9.4 and
−8.3‰. Where surface fluxes were statistically significant, changes in δ13CCO2 between samples taken
at times 0 and 45 min were used to calculate the δ13CΔCO2 of the surface flux (4.5). The δ13CΔCO2 of
significant positive and negative surface CO2 fluxes was highly variable, ranging from −2.4 to −31.8‰
(Figure 4.3B). The δ13CΔCO2 of the highest effluxes ranged between −2.4 and −8.0‰, whereas the
highest influxes had δ13CΔCO2 values of −14.4 and −12.7‰. The most negative δ13CΔCO2 values of −23.4
to −31.8‰ were measured following the four-hourly intervals during which a transition from positive
to negative (or vice versa) fluxes occurred. The measured and predicted δ13CΔCO2 values were
generally in good agreement (Figure 4.3B).
Subsurface CO2 profiles and fluxes
Values for CO2 concentration and δ13CCO2 at 0 cm depth were relatively consistent throughout the 48h sampling period (Figure 4.4). Despite these samples being taken 1 cm above the ground surface,
they are assumed to provide the soil−atmosphere boundary CO2 concentration and δ13CCO2 over the
sampling period, averaging 400 ± 4 µL L-1, and −8.8 ± 0.2‰, respectively. These values differ slightly
from atmospheric CO2 concentration (385 µL L-1) and δ13CCO2 (−8.2‰) measured at the South Pole in
December 2009 (Keeling et al., 2001).
Temporal changes in CO2 concentration and δ13CCO2 profiles throughout the two 24-h sampling
periods at Site B were generally symmetrical (Figure 4.4). The CO2 concentration profiles were quasisymmetrical about a vertical line corresponding to the local atmospheric CO2 concentration (400 µL
L-1), with maximum variation in concentration ranging from 372 to 426 µL L-1 between 10 and 15 cm
depth. Subsurface soil CO2 concentrations were generally greater than atmospheric CO2
concentration at 10:30, 14:30 and 18:30 h, and less than atmospheric CO2 concentration at 22:30,
02:30 and 06:30 h.
Depth profiles of δ13CCO2 were quasi-symmetrical about a line sloping from −8.8‰ at the surface to
−8.3‰ at 30 cm depth for both 24-h periods (Figure 4.4). Changes in δ13CCO2 values tended to
correspond to changes in CO2 concentration, with δ13CCO2 generally becoming more depleted as CO2
concentration decreased, and more enriched as CO2 concentration increased. The maximum
variation in δ13CCO2 occurred at 5 cm depth, with δ13CCO2 ranging between −9.1 and −8.3‰, whereas
the minimum variation was at 15 cm depth.
74
Figure 4.4
Soil profile CO2 concentrations and δ13CCO2 values at Site B, Taylor Valley. (A) and (B)
represent data from Day 1 and Day 2, respectively. Data points are averages from each
sampling time (n = 1 or 2 for 0 cm samples. At all other depths, n = 4). Error bars
represent the standard error of the mean.
75
Subsurface storage CO2 fluxes, calculated based on the change in the profile average CO2
concentration between sampling times (equation 4.7), varied diurnally between positive and
negative values, with the same periodicity as surface CO2 fluxes. However, they were an order of
magnitude lower, ranging from −0.01 to 0.01 µmol m-2 s-1 (Figure 4.3A). Changes in the direction of
subsurface storage CO2 fluxes preceded changes in the direction of surface fluxes by 4−6 h.
4.3.3 Linking surface and subsurface CO2 dynamics
Day 1 (Figure 4.3 and Figure 4.4A)
14:30 – 18:30 h
Between 14:30 and 18:30 h, the soil was in a steady-state phase of net CO2 efflux. Subsoil CO2
concentrations were above ambient, and increased only slightly below 5 cm depth during this
interval. The positive concentration gradients (increasing CO2 concentration with depth) drove
positive surface CO2 fluxes, which remained close to the maximum rate of 0.058 µmol m-2 s-1 over this
interval. Subsurface δ13CCO2 values did not change significantly, and were close to their most enriched
(−8.0 to −8.5‰).
Subsurface δ13CCO2 gradients were also positive (increasing δ13CCO2 with depth), which, together with
the positive CO2 concentration gradients, drove surface fluxes enriched in 13C relative to the ambient
soil atmosphere (~ −8.8‰). The δ13CΔCO2 of the measured flux into surface chambers was −2.4 and
−5.1‰ at 14:30 and 18:30 h, respectively; the δ13CΔCO2 of fluxes calculated from CO2 and δ13CCO2
gradients at those times were 3.2 and −5.2‰, respectively. Over this period, soil temperature at 5
cm depth remained relatively constant and close to its daily maximum, and soil temperature at 15 cm
depth increased linearly at close to its maximum rate of change.
18:30 – 22:30 h
Between 18:30 and 22:30 h, the soil entered a transient phase of CO2 loss. Subsurface CO2
concentrations decreased throughout the profile, most markedly between 5 and 15 cm depth, and by
22:30 h, the positive concentration gradient that existed between 0 and 10 cm depth at 18:30 h had
shifted to a negative concentration gradient (decreasing with increasing depth) between 0 and 5 cm
depth, with CO2 concentrations above 15 cm depth all lower than that of atmospheric CO2. Despite
the efflux of CO2 at 18:30 h being enriched in 13C relative to ambient soil CO2, the depth profile of
δ13CCO2 changed very little over the 4-h period, and the positive δ13CCO2 gradient was maintained.
Surface CO2 fluxes shifted from positive to negative (0.054 to −0.026 µmol m-2 s-1) in response to
changes in the CO2 concentration gradient. The switch to a negative CO2 concentration gradient,
combined with the persistent positive δ13CCO2 gradient, drove a CO2 influx depleted in 13C. The
δ13CΔCO2 of CO2 influxes calculated from surface chambers averaged −23.4‰ at 22:30 h, which was
consistent with the δ13CΔCO2 predicted from CO2 and δ13CCO2 gradients of −26.9‰.
76
The marked decrease in CO2 concentration throughout the profile coincided with rapidly decreasing
soil temperature at 5 cm depth, which shifted from increasing at close to its maximum positive rate
to reach its daily maximum early in the interval, to decreasing at its maximum negative rate. Soil
temperature at 15 cm depth increased over this period (albeit at a slower rate than during the
previous 4-h period), reaching its daily maximum at the end of the interval.
22:30 – 02:30 h
Over the period from 22:30 to 02:30 h, the soil remained in a transient phase of CO2 loss, although
the rate of CO2 absorption was lower than during the previous 4-h period. Subsurface CO2
concentration decreased below 5 cm depth, and the negative concentration gradient between 0 and
5 cm depth at 22:30 h extended to 10 cm depth by 02:30 h, with soil CO2 concentration being less
than ambient atmospheric CO2 concentration throughout the profile. Subsurface δ13CCO2 values
became significantly more depleted (by up to −0.5‰) in the upper 15 cm of the profile, consistent
with the highly depleted surface influx established at 22:30 h, and resulting in a weak positive δ13CCO2
gradient. Surface CO2 fluxes became increasingly negative, consistent with the steeper negative
concentration gradient in the subsoil, and reached a daily maximum (negative) rate of −0.039 µmol
m-2 s-1 at the end of the interval. The average δ13CΔCO2 of the CO2 influx shifted from −23.4‰ at 22:30
h to −14.4‰ at 02:30 h, which was again consistent with predicted δ13CΔCO2 values of the net surface
CO2 influx (−26.9 and −17.3‰ at 22:30 and 02:30 h, respectively). The rates of change in soil
temperature at 5 and 15 cm depth were both negative, with the latter at its maximum negative rate
by the end of the interval.
02:30 – 06:30 h
During this period, the soil was in a steady-state phase of net CO2 influx. Subsurface soil CO2
concentration and δ13CCO2 depth profiles did not change significantly, despite there being CO2 fluxes
into the soil of −0.039 µmol m-2 s-1 at 02:30 h and −0.029 µmol m-2 s-1 at 06:30 h. This set of
circumstances requires that the combination of biotic production (if any), abiotic consumption, and
diffusion associated with the subsoil CO2 concentration gradients have balanced effects with respect
to the total mass of CO2 and abundances of 12CO2 and 13CO2.
However, subtle changes in δ13CCO2 values at 5 cm depth, owing to the δ13CΔCO2 of the CO2 influx at
02:30 h being isotopically lighter than ambient soil CO2, acted to change the δ13CCO2 gradient from a
weak positive gradient at 02:30 h to a weak negative gradient at 06:30 h. This, combined with the
negative CO2 concentration gradient throughout the interval, drove a heavier influx of CO2 at 06:30 h.
The δ13CΔCO2 of the surface CO2 influx shifted from −14.4‰ at 02:30 h to −5.6‰ at 06:30 h; these
values were similar to the predicted δ13CΔCO2 values of −17.3 and −10.5‰ at 02:30 and 06:30 h,
respectively. Soil temperature at 5 cm depth remained close to its daily minimum throughout the
77
interval, while soil temperature at 15 cm depth continued to decrease at close to its maximum
negative rate. These temperature changes mirrored the soil temperature conditions that prevailed
during the period of steady-state efflux (14:30 – 18:30 h).
06:30 – 10:30 h
Between 06:30 and 10:30 h, the soil entered a transient phase of CO2 gain. Subsurface CO2
concentrations increased significantly at all depths, at a rate which was close to the maximum
subsurface accumulation rate. The negative CO2 concentration gradient between 0 and 15 cm depth
at 06:30 h switched to a slight positive gradient between 0 and 5 cm depth at 10:30 h; this was
reflected in surface CO2 fluxes which shifted from negative to positive over this period. Subsurface
δ13CCO2 values became slightly more depleted in the upper 10 cm of the profile and slightly more
enriched in the lower 10 cm, thus increasing the negative δ13CCO2 gradient. The change in CO2
concentration gradient from negative to positive, combined with the persistent negative δ13CCO2
gradient drove a highly depleted (−26.1‰) surface CO2 efflux at 10:30 h. This value, although being
very close to the δ13C value of organic matter measured in the top 5 cm of the soil profile at Site B
(Chapter 3, Figure 3.8C), is relatively close to that predicted by CO2 concentration and δ13CCO2
gradients (−19.1‰). Soil temperature at 5 cm depth increased from close to its daily minimum, and
soil temperature at 15 cm depth decreased at a low rate, reaching its daily minimum by the end of
the interval.
10:30 – 14:30 h
This interval completes a full 24-h cycle, thus data from the subsurface profile measured at 14:30 h
are shown on the “Day 2” profile (Figure 4.4B). As for the Day 1 22:30 to 02:30 h period of subsoil
consumption of CO2 and negative surface fluxes, this interval also marks significant change in both
CO2 concentration and δ13CCO2 profiles. Accumulation of CO2 in the subsoil continued from the
previous 4-h interval, and by 14:30 h, subsurface CO2 concentrations were greater than atmospheric
CO2 concentration throughout the profile, with the positive CO2 concentration gradient extending to
15 cm depth. The average surface CO2 flux rate increased from 0.029 µmol m-2 s-1 at 10:30 h to the
daily maximum rate of 0.057 µmol m-2 s-1 at 14:30 h, which is very similar to the maximum surface
flux rate at 14:30 h on Day 1. Soil δ13CCO2 became more enriched throughout the profile, consistent
with the isotopically light efflux of CO2 at 10:30 h acting to increase the abundance of the heavier 13C
isotope in the soil atmosphere, with a maximum shift from −9.1 to −8.3‰ at 5 cm depth. The
enrichment in soil δ13CCO2 acted to change the δ13CCO2 gradient from negative to positive, and
combined with the positive concentration gradient, drove a relatively heavy (−8.0‰) efflux of CO2 at
14:30 h. This is close to the δ13CΔCO2 of the CO2 efflux (−4.5‰) predicted from the CO2 concentration
and δ13CCO2 gradients at 14:30 h. Over this interval, soil temperature at both 5 and 15 cm depth
78
increased; temperature at 5 cm depth increased at its maximum rate, and soil temperature at 15 cm
depth increased from its daily minimum.
Day 2 (Figure 4.3 and Figure 4.4B)
14:30 – 18:30 h
Unlike the steady-state condition that prevailed between 14:30 and 18:30 h on Day 1, during this
interval, the subsoil CO2 concentration decreased throughout the upper 20 cm of the profile,
although the δ13CCO2 changed very little. By 18:30 h, the positive concentration gradient that existed
between 0 and 15 cm depth at 14:30 h had shifted to a negative concentration gradient between 0
and 5 cm depth, and the profile average CO2 concentration was close to that of atmospheric CO2. The
δ13CCO2 profiles did not change significantly in the upper 15 cm of the soil, as the efflux of CO2 at
14:30 h was of similar isotopic composition to that of soil CO2; below this depth, δ13CCO2 became
slightly more depleted. The average surface CO2 flux rate decreased from its daily maximum positive
rate to near zero in response to changes in the CO2 concentration gradient. The switch to a negative
concentration gradient, combined with the persistent positive δ13CCO2 gradient acted to drive a
relatively depleted flux at 18:30 h, with the average δ13CΔCO2 of the net flux shifting from −8.0‰ at
14:30 h to −31.8‰ at 18:30 h. The δ13CΔCO2 values of the net fluxes were consistent with predicted
δ13CΔCO2 values of −4.5 and −30.0‰ at 14:30 and 18:30 h, respectively. Soil temperatures at 5 and 15
cm depth reached their daily maxima during this interval. However, soil temperature at 5 cm depth
predominantly decreased, at times at a rate close to its maximum negative rate of change, whilst soil
temperature at 15 cm depth increased to its daily maximum at its maximum rate of change. These
temperature changes, along with the subsurface absorption of CO2, decrease in surface CO2 flux rate,
and decrease in subsurface CO2 concentration with little change in subsurface δ13CCO2 over this
interval were very similar to those which occurred during the 18:30 to 22:30 h interval on Day 1.
18:30 – 22:30 h
Between 18:30 and 22:30 h, the soil continued to absorb CO2, at a rate slightly higher than during the
previous interval. Subsurface CO2 concentrations decreased at all depths, and the negative
concentration gradient between 0 and 5 cm depth at 18:30 h extended to 10 cm depth by 22:30 h,
with CO2 concentrations throughout the profile being less than that of atmospheric CO2. Subsurface
δ13CCO2 values became significantly more depleted (by up to −0.4‰) throughout the profile. This
change was consistent with a highly depleted CO2 influx established by the CO2 and δ13CCO2 gradients
at 18:30 h, and resulted in a weak positive δ13CCO2 gradient. The average surface CO2 flux rate became
negative, and at 22:30 h, the δ13CΔCO2 of the CO2 influx was −12.9‰, which was consistent with the
predicted δ13CΔCO2 of −17.5‰. Soil temperatures decreased at both 5 and 15 cm depth. The soil
temperature trends and changes in CO2 dynamics observed during this period were similar to those
occurring between 22:30 and 02:30 h on Day 1.
79
22:30 – 02:30 h
This period represents the closest approximation to a steady-state period of CO2 influx on Day 2.
During this period, the soil continued to absorb CO2. Subsurface CO2 concentrations continued to
decrease below 5 cm depth, and the negative concentration gradient extended to 15 cm depth,
although the subsurface δ13CCO2 profile did not significantly change. The average surface CO2 flux rate
continued to become increasingly negative, reaching its maximum negative rate of −0.041 µmol m-2 s1
by 02:30 h. The average δ13CΔCO2 of the CO2 influx remained very similar at −12.7‰, which was again
consistent with the δ13CΔCO2 of the CO2 influx (−17.7‰) predicted from CO2 concentration and δ13CCO2
gradients at 02:30 h. As the δ13CCO2 profiles did not change significantly, this implies that the CO2
being absorbed during this period had a δ13CCO2 value close to that of the CO2 influx over this period
(−12.9‰). Soil temperatures at 5 and 15 cm depth continued to decrease; the rate of change in soil
temperature at 15 cm depth was close to its maximum negative rate. The decrease in subsurface CO2
concentrations and surface flux rates was similar to the trends occurring during the 22:30 to 02:30 h
interval on Day 1, with the main difference during this interval on Day 2 being that subsurface δ13CCO2
values did not significantly change.
02:30 – 06:30 h
Over this interval, the subsoil returned to a transient CO2 accumulation phase. However, the net
accumulation was small; CO2 concentrations remained below that of atmospheric throughout the
profile, and the negative concentration gradient between 0 and 15 cm depth was maintained.
Subsurface δ13CCO2 values became significantly more depleted (by up to −0.5‰) in the upper 15 cm,
with the maximum depletion in δ13CCO2 occurring at 5 cm depth. The depletion in δ13CCO2 is consistent
with an influx of CO2 isotopically lighter than that of the soil atmosphere at 02:30 h, and resulted in a
negative δ13CCO2 gradient at 06:30 h. The average surface CO2 flux rate increased from the maximum
negative rate at 02:30 h to a moderate positive rate by 06:30 h. The positive CO2 flux at 06:30 h is
inconsistent with the negative CO2 concentration gradient that persisted throughout this interval.
Additionally, the average δ13CΔCO2 of the net CO2 efflux at 06:30 h of −28.9‰ is inconsistent with the
combination of CO2 concentration and δ13CCO2 gradients at that time, which predict a CO2 influx at
13.9‰. The surface flux and its isotopic composition at 06:30 h are the only anomalous data points
recorded throughout the 48-h sampling period.
Soil temperature at 5 cm depth reached its daily minimum during this interval, and soil temperature
at 15 cm depth decreased at close to its maximum negative rate, reaching its daily minimum at the
end of the interval. The subsurface CO2 concentration profiles were similar to those observed during
the 02:30 to 06:30 h interval on Day 1. However, during the Day 2 interval, subsurface δ13CCO2 values
became significantly more depleted in the upper 15 cm of the profile, which did not occur during the
02:30 to 06:30 h interval on Day 1. The increase in surface CO2 flux rate from a negative to a positive
80
rate, and the changes in δ13CΔCO2 of the net CO2 exchanged within surface chambers from enriched to
highly depleted values, along with changes in soil temperature, were more similar to those observed
during the 06:30 to 10:30 h interval on Day 1.
06:30 – 10:30 h
From 06:30 h to the final sampling at 10:30 h, the soil continued its transient accumulation of CO2 at
its maximum rate. Subsurface CO2 concentration increased throughout the profile, and by 10:30 h,
the negative concentration gradient between 0 and 15 cm depth had shifted to a positive
concentration gradient between 0 and 5 cm depth, with CO2 concentrations throughout the profile
being greater than that of atmospheric CO2. Despite the significant increase in subsurface CO2
concentrations, δ13CCO2 values did not significantly change, and the δ13CCO2 gradient remained
negative. The average surface CO2 flux rate continued to increase, and at 10:30 h, the average
δ13CΔCO2 of the net CO2 efflux was −12.3‰, which was close to the predicted δ13CΔCO2 value of
−15.2‰. Soil temperature at 5 cm depth increased, and soil temperature at 15 cm depth increased
from its daily minimum. The increases in surface CO2 flux rate and increases in soil temperature
observed over this period were similar to those that occurred during the 10:30 to 14:30 h interval on
Day 1.
Summary
The symmetrical patterns in CO2 concentration and δ13CCO2 profiles exemplified by data from Day 1
highlight important periods of steady-state and transient CO2 dynamics. At the warmest time of the
day (14:30–18:30 h), the system was in steady-state. Subsurface CO2 concentrations were high,
positive concentration gradients drove surface effluxes, and the δ13CCO2 of soil CO2 was relatively
enriched, with subsurface δ13CCO2 values greater than that of atmospheric CO2. As the soil began to
cool, it entered a transient phase during which soil CO2 concentrations decreased, such that by 22:30
h, the CO2 concentration gradient was negative, whilst the δ13CCO2 profile remained unchanged. This
resulted in a surface influx of relatively depleted CO2. By 02:30 h, soil CO2 concentrations had
reached a minimum. The depleted surface CO2 influxes shifted the subsurface δ13CCO2 to more
negative values. The relatively depleted δ13CCO2 profile, together with the low CO2 concentrations,
persisted in a phase of steady-state influx until 06:30 h. As the soil began to warm, another transient
phase ensued. This was characterised by increased soil CO2 concentrations and a shift back to a
positive concentration gradient and surface efflux by 10:30 h, whilst the δ13CCO2 profile remained
unchanged. The switch in CO2 concentration gradient from negative to positive, combined with an
unchanged δ13CCO2 profile drove a highly depleted surface efflux which acted to shift the soil δ13CCO2
to more enriched values. By 14:30 h (the beginning of sampling on Day 2), the soil once again had
relatively high CO2 concentration and δ13CCO2 values. Similar behaviour occurred over Day 2, but
81
subtle differences in soil temperature variability meant that the sampling periods captured different
stages of the transient CO2 dynamics.
4.4 Discussion
4.4.1 Diel variability in surface and subsurface storage CO2 flux rates
The relatively small surface CO2 fluxes measured throughout the 48-h sampling period are consistent
with in situ surface flux rates measured in Taylor Valley by Burkins et al. (2001) and Parsons et al.
(2004), as well as results presented in Chapter 3. However, they are an order of magnitude lower
than in situ surface CO2 flux rates measured by Ball et al. (2009). All studies utilised sites with similar
soil pH, EC, organic and inorganic C values; soil moisture content and the magnitude of diel
temperature variability were also comparable. Despite the relatively high flux rates measured by Ball
et al. (2009), the variability in surface CO2 fluxes seen in their study and by Parsons et al. (2004)
showed similar diel patterns to those observed in this study, with positive and negative surface fluxes
corresponding to periods when soil temperatures were increasing, and decreasing, respectively.
Surface CO2 fluxes were most strongly correlated with air temperature, measured 2 cm above the
soil surface (R2 = 0.96; p = <0.001), and less so with soil temperature at 5 cm depth (R2 = 0.78; p =
<0.001; Figure 4.5). Aside from the correlation with air temperature, the best explanatory variable
for the magnitude of surface flux rates was the rate of change in soil temperature at 15 cm depth (R2
= 0.85; p = <0.001; Figure 4.6). The rate of change in soil temperature at 5 cm depth only explained
43% of flux variability (p = 0.02).
The strong correlation between surface CO2 flux rate and air temperature is difficult to interpret, and
there may in fact be no direct causative link. However, the strong correlation between surface CO2
flux rate and ΔT/Δt at 15 cm depth is consistent with surface CO2 fluxes being driven by temperaturecontrolled solubility of CO2 in soil solution, according to Henry’s Law, as posited by Parsons et al.
(2004) and Ball et al. (2009). The greater influence of soil temperature changes at 15 cm depth
relative to 5 cm is surprising, as the amplitude of temperature variability is greater at 5 cm depth,
and soil moisture content is similar at both depths. The correlation between surface CO2 flux rate and
the rate of change in soil temperature at 15 cm depth in this study contrasts with the strong
correlation between surface CO2 flux rates and the rate of change in soil temperature at 5 cm depth
shown by Parsons et al. (2004), although their study did not include soil temperature measurements
at any other depths.
82
Figure 4.5
Relationship between average temperature over the surface flux sampling period and
surface CO2 flux rates at Site B, Taylor Valley.
Figure 4.6
Relationship between the average rate of change in soil temperature over the surface
flux sampling period at 5 cm and 15 cm depth with surface CO2 flux rates at Site B, Taylor
Valley.
Whereas Ball et al. (2009) inferred from their heat sterilised in situ microcosm study that a biological
CO2 flux was superimposed on the abiotic temperature-related fluxes, cumulative 24-hr CO2 fluxes in
83
this study detect no statistically significant net flux over six successive 24-hr periods (Figure 4.7), and
hence suggest a purely abiotically-driven system, where influxes and effluxes balance. Nonetheless,
the presence of a biotic component to the flux must be considered. This is explored further below in
relation to the δ13CCO2 of the soil atmosphere and surface CO2 fluxes. Regardless of the significance of
a biological component to the total CO2 flux, abiotic processes must be invoked – at least to explain
CO2 consumption beneath the opaque surface chambers used in this study, which precluded
photosynthetic uptake of CO2.
Figure 4.7
Net cumulative surface CO2 fluxes at Site B, Taylor Valley, over consecutive 24-h periods
beginning at 14:30 h on Day 1. Maximum and minimum net fluxes represent
uncertainties calculated from the standard error of the mean.
In Chapter 3, temperature-controlled exsolution and dissolution of CO2, according to Henry’s Law,
were modelled analytically and compared to measured changes in subsurface soil CO2
concentrations, as a test of the hypothesis that this phenomenon was the main driver of CO2
dynamics. However, this approach was predicated on a closed-system approximation. In this study,
the relatively high frequency at which subsurface CO2 concentration measurements were made
enables subsurface storage fluxes (fluxes that result in changes in CO2 storage in the soil) to be
estimated over four-hourly intervals. Results show the amplitude of surface fluxes is more than five
times larger than storage fluxes, and hence a closed system assumption is not appropriate, at least
84
for near-surface samples. However, the storage fluxes themselves provide another test of the
hypothesis, as they reflect the production and consumption of CO2 in the soil, and can be considered
alongside soil temperature changes.
Like surface fluxes, storage fluxes also showed diel variation (Figure 4.3A), switching from positive to
negative as CO2 concentration within the profile increased and decreased, respectively. This
variability led surface flux variations by 4−6 hours. Storage fluxes were explained by the rate of
change in soil temperature at 5 cm depth (R2 = 0.84; p = <0.001; Figure 4.8), but showed no
relationship to the rate of change in soil temperature at 15 cm depth. This is consistent with the
suggestion that Henry’s Law-driven dissolution and exsolution of CO2 drives changes in subsurface
CO2 concentrations, and indicates that, as expected, the shallower depths of the soil (where diel
temperature changes are greatest) are the most important for determining the subsurface CO2
concentration changes that ultimately drive surface fluxes. The better correlation between surface
CO2 flux rates and the rate of change in soil temperature at 15 cm depth compared to 5 cm depth
discussed above is most likely an artefact. Lags in warming and cooling of the subsoil relative to
shallower depths, and the lag between subsurface CO2 production or consumption and changes in
surface CO2 flux may bring about coincidence of their behaviours and promote the strong
correlation.
Figure 4.8
Relationship between the rate of change in soil temperature at 5 cm and 15 cm depth
with subsurface storage CO2 flux rates at Site B, Taylor Valley. The solid black lines
represent linear relationships with equations as follows. At 5 cm depth, y = 0.0059x +
0.0002. At 15 cm depth, y = 0.0047x - 0.0001.
85
4.4.2 Soil gas transport
The significant linear relationship between subsurface CO2 concentration gradient and surface CO2
flux rate (R2 = 0.91, p <0.001; Figure 4.9) is strong evidence that molecular diffusion processes, as
described by Fick’s Law, dominate CO2 fluxes. The slope of the line approximates the effective
diffusivity, De. However, advection caused by wind-induced pressure fluctuations (termed "wind
pumping" or "pressure pumping"; Fukuda, 1955; Massman et al., 1997), which can be further
modified by interactions with surface topography (Colbeck, 1989; Clarke and Waddington, 1991),
must be considered. Wind pumping acts to enhance the rate of diffusive gas transport along a given
concentration gradient, thereby producing flux rates greater than those expected from molecular
diffusion alone (Bowling and Massman, 2011). This enhancement may be represented
mathematically as an additive component to the effective diffusivity in a Fick’s Law representation of
gas transport:
𝑄𝑄 = (𝐷𝐷𝑒𝑒 + 𝐾𝐾)
𝑑𝑑𝑑𝑑
,
𝑑𝑑𝑑𝑑
(4.27)
where Q is the gas flux, De is the effective molecular diffusivity, K is a wind-dependent enhanced
diffusivity parameter (Bowling and Massman, 2011), and dC/dz is the gas concentration gradient.
Figure 4.9
Relationship between subsurface CO2 concentration gradient and surface CO2 flux on
combined data from Days 1 and 2, Site B, Taylor Valley. The solid black line represents a
linear relationship: y = 0.032x + 8E-07; the slope of the line represents the effective
diffusivity of CO2 in the soil. Anomalous data from 06:30 h on Day 2 were excluded.
86
Therefore, the slope of the line in Figure 4.9 may be an overestimate of the true effective molecular
diffusivity if it incorporates a wind pumping enhancement. Takle et al. (2004) found that under
conditions conducive to wind pumping, surface soil CO2 flux rates were up to ten times greater than
those expected from molecular diffusion alone. Similarly, in a study of CO2 transport in a mountain
forest snowpack, Bowling and Massman (2011) found that wind pumping enhanced transport of CO2
by up to 40% in the short term (hours). In this study, wind speed was measured 1.2 m above the soil
surface at Site A, located 2.8 km east of Site B. The strong diel pattern in wind strength is consistent
with an anabatic easterly wind blowing up-valley during the warmest parts of the day. It is almost
certain that the same wind blew at Site B; at the very least, the exposed nature of the site means
that wind speeds at Site B were unlikely to have been lower than those measured at Site A.
If wind pumping significantly enhanced the surface CO2 flux rate for a given concentration gradient,
then positive fluxes during the warmest parts of the day, when the wind was strongest (~ 4 m s−1),
would be greater than negative fluxes, which occurred during colder parts of the day when wind
speeds were lower. This would manifest in a plot of concentration gradient versus surface CO2 flux
rate (Figure 4.9) as distinct linear relationships within the positive (steeper gradient) and negative
(shallower gradient) flux domains. If such data were fitted with a single line, that line would have a
significant, greater-than-zero, intercept. The intercept in Figure 4.9 is not significantly different from
zero (p = 0.067). Further evidence of the relative unimportance of a wind pumping effect at this site
is seen in a plot of wind speed versus surface CO2 flux rate (Figure 4.10), which, despite the
significant relationship (p = 0.027), accounts for only 40% of the variation in flux rate.
Figure 4.10 Relationship between wind speed and surface CO2 flux rates at Site B, Taylor Valley. The
solid black line represents a linear relationship: y = 0.0189x − 0.0353.
87
The apparent insignificance of wind pumping in this study is in contrast to Risk et al.’s (2013) study of
CO2 fluxes in the Dry Valleys, which found that measured CO2 fluxes were best related to average
wind speeds in the 16 days prior to flux measurements. However, given that CO2 fluxes in Risk et al.’s
(2013) year-long study were only sampled every 4 days during winter, and that Antarctic dry valley
soils are characteristically dry and porous (McCraw, 1967; Campbell et al., 1997), it seems unlikely
that any wind pumping due to strong winds over the previous 16 days would influence surface CO2
fluxes. Any enhancement of surface CO2 fluxes by wind pumping would likely be related to wind
speed and associated changes in pressure at the time of sampling (e.g. Takle et al., 2004). However,
differences in the frequency of sampling and the period over which data were obtained between this
study and Risk et al.’s (2013) study means that there is not a good basis for comparing the datasets.
The time-series of flux measurements presented in this study is too short to investigate long lag-time
effects.
Therefore, assuming the slope of the line in Figure 4.9 is a reliable estimate of the effective
(molecular) diffusivity De, and that the low soil moisture contents had a negligible influence on CO2
diffusivity in the liquid phase (Fang and Moncrieff, 1999), particularly in this very dry soil, the
relationship
𝐷𝐷𝑒𝑒 = 𝐷𝐷 ∙ 𝜀𝜀 ∙ 𝜏𝜏 ,
(Massman, 2006)
(4.28)
where D is the free air diffusivity, and ε is the air-filled soil porosity, can be used to estimate the
tortuosity factor, τ. Using a free air diffusivity for CO2 (at 1°C and atmospheric pressure) of 0.140 cm2
s−1 (Marrero and Mason, 1972), a soil porosity of 0.4 (from empirical data), a soil moisture content of
0.01, and a De of 0.032 ± 0.003 cm2 s−1 yields a tortuosity value of 1.8 ± 0.2. This value is in good
agreement with the range of tortuosity values (1.5 – 2.0) calculated by McKay et al. (1998) in their
gas diffusion experiments on Dry Valley soils, and lends further support to the inference that wind
pumping was not a significant modulator of CO2 fluxes in this study. A wind pumping effect
contributing to an overestimate of De would result in an unexpectedly large tortuosity factor.
In addition, the effective diffusivity determined from Figure 4.9 is in reasonable agreement with that
estimated from the Millington−Quirk equation which is regarded as the best-performing soil typeindependent model of gas diffusivity (Moldrup et al., 2004). A more advanced soil type-dependent
prediction of gas diffusivity was not determined, owing to the unavailability of soil water
characteristic data. Furthermore, the extremely low soil moisture content at this site meant that
McCarthy and Johnson’s (1995) modified version of the Millington−Quirk equation, which includes
the effects of aqueous-phase diffusion, was not required.
88
4.4.3 Sources of CO2
This study, for the first time in Antarctica, used stable C isotope ratios measured from subsurface soil
CO2 and surface CO2 fluxes in an attempt to trace the sources of CO2 in Dry Valley soils. Carbon stable
isotope studies have been commonly used in temperate environments to distinguish between soiland root-respired CO2 in partitioning experiments that rely on the different isotopic composition of
metabolised C sources (e.g. Rochette et al., 1999; Ekblad and Högberg, 2000), or in pulse labelling
experiments where the soil is flooded with exotic CO2 of distinctive (extreme) δ13C value. As the
potential sources of C in Dry Valley soils are limited to highly isotopically depleted organic C (Burkins
et al., 2000; Hopkins et al., 2009) and relatively enriched inorganic C (Foley, 2005), it would appear
that the two sources could be readily discriminated by using an isotopic approach.
However, characterising the isotopic composition of soil-produced CO2 requires that the system is in
steady state (Amundson et al., 1998). Under this condition, and only then, does the isotopic
composition of a surface CO2 flux match that of the CO2 being produced in the soil. Under non-steady
conditions, measured surface CO2 flux rates do not represent CO2 production at that time, and the
δ13CCO2 of the surface flux does not represent the true value of the δ13CCO2 being produced (Nickerson
and Risk, 2009c). Instead, even if the δ13C of the CO2 produced is constant, transient changes in the
CO2 production rate, driven by transient changes in environmental variables, combine with
differences in the diffusivities of 12CO2 and 13CO2 to produce different concentration gradients for the
two isotopologues. This induces a kinetic fractionation effect, and consequently, the isotopic
composition of the surface CO2 flux is different to that of the CO2 being produced in the soil. The
term ‘dynamic fractionation’ was coined by Nickerson and Risk (2009c) to describe this phenomenon.
Consequently, under conditions of dynamic fractionation, simple isotope mixing models cannot be
used to partition contributions from different sources to a surface flux. The influence of dynamic
fractionation on the δ13C of surface CO2 fluxes was highlighted by Moyes et al. (2010) in their study of
variability in heterotrophic and rhizosphere sources of soil respiration under deciduous trees. During
periods of low flux, they observed large diel variability in the δ13C of surface CO2 fluxes (up to 12‰).
However, by modelling diffusive transport of 12CO2 and 13CO2 in the soil, under conditions of a
constant δ13C of respiratory production, they found that diel variability in the δ13C of surface fluxes
could be explained by dynamic fractionation, rather than requiring any change in the source
substrate.
In this study, dynamic fractionation effects combined with switches from positive to negative CO2
fluxes lead to extreme changes in the δ13CCO2 of surface CO2 fluxes. The isotopic composition of
measured surface CO2 fluxes ranged between −2.4 and −31.8‰ (Figure 4.3B), the greatest variability
yet reported globally, while the isotopic composition of subsurface soil CO2 varied minimally,
89
between −7.9 and −9.1‰ (Figure 4.4). The greatest shifts in the δ13CCO2 of the surface flux occurred
when fluxes were either rapidly increasing or decreasing, resulting from shifts in behaviour from CO2
consumption to production, or vice versa, respectively. At these times, subsurface CO2 concentration
gradients typically changed quickly, while δ13CCO2 gradients remained relatively unchanged.
Consequently, subsurface CO2 concentration and δ13CCO2 gradients typically switched from having the
same sign (positive or negative; Figure 4.11A and D), to being of opposite sign (Figure 4.11B and C).
As gradients of δ13CCO2 act in conjunction with CO2 concentration gradients to influence the isotopic
composition of the net efflux or influx of CO2, any change in the sign of either gradient has a
corresponding influence on the isotopic composition of the surface CO2 flux, which subsequently
influences the isotopic composition of the subsurface soil CO2, as illustrated in the schematics below
(Figure 4.11).
Figure 4.11 Schematics showing the influence of CO2 concentration and δ13CCO2 gradients on the
δ13CΔCO2 of surface CO2 fluxes and on the δ13CCO2 of soil CO2. The letters A, B, C, and D
refer to various scenarios described in the text.
90
A positive δ13CCO2 gradient (increasing δ13CCO2 with increasing depth) will act to preferentially drive
13
CO2 out of the profile when the CO2 concentration gradient is positive (Figure 4.11A), whereas
when the CO2 concentration gradient is negative, the same positive δ13CCO2 gradient will act to drive a
preferential influx of 12CO2 (Figure 4.11B). The transition from a positive to a negative concentration
gradient, without a corresponding change in the δ13CCO2 gradient, commonly occurred following a
switch from a period of soil warming to soil cooling, and resulted in a relatively light CO2 influx
(Figure 4.11B). Similarly, a negative δ13CCO2 gradient (decreasing δ13CCO2 with increasing depth) will act
to preferentially drive 12CO2 out of the profile when the CO2 concentration gradient is positive (Figure
4.11C), and the same negative δ13CCO2 gradient will drive a preferential influx of 13CO2 when the CO2
concentration gradient is negative (Figure 4.11D). Following a switch from a period of soil cooling to
soil warming, the subsurface CO2 concentration gradient typically switched from negative to positive,
whilst the δ13CCO2 gradient remained relatively unchanged. This resulted in a relatively light CO2 efflux
(Figure 4.11C).
In addition, the δ13CΔCO2 of surface CO2 fluxes has a feedback effect on subsurface δ13CCO2 values. At
times when CO2 concentration and δ13CCO2 gradients drive a heavy efflux (Figure 4.11A) or light influx
of CO2 (Figure 4.11B), the isotopic composition of the soil CO2 will become lighter. Conversely, at
times when there is a light efflux (Figure 4.11C) or heavy influx of CO2 (Figure 4.11D), the soil CO2 will
become isotopically heavier. Changes in subsurface δ13CCO2 values were consistent with the direction
and isotopic composition of the surface flux. As such, it is likely that changes in subsurface δ13CCO2
values resulted from the large shifts in the δ13CΔCO2 of surface CO2 fluxes, rather than any major
change in the δ13CCO2 of CO2 being produced or consumed.
The high diel variability in surface CO2 flux rates in Dry Valley soils highlights their characteristically
non-steady nature, and, as variability in the rate of CO2 production has the potential to perpetuate
non-steady conditions, dynamic fractionation is an important factor to consider in understanding the
δ13CΔCO2 of surface fluxes. The strong correlation (R2 = 0.80; p = <0.001) between measured δ13CΔCO2
values and those predicted based on CO2 concentration and δ13CCO2 gradients (Figure 4.12) suggests
that the relatively coarse resolution (5 cm) at which CO2 concentration and δ13CCO2 profiles were
sampled was sufficient to capture the zones of CO2 production and consumption.
91
Figure 4.12 Relationship between the δ13CΔCO2 of measured and modelled surface CO2 fluxes from
Days 1 and 2, Site B, Taylor Valley. The anomalous data point from 06:30 h on Day 2 was
excluded from this plot and from the linear regression. The solid black line is a linear
relationship relating x (measured) and y (modelled) values as follows: y = 0.94x – 1.42.
The only occasions when the δ13CΔCO2 of surface CO2 fluxes could be related to the isotopic
composition of CO2 produced or consumed within the soil was during three phases that
approximated steady-state with respect to CO2 concentration and δ13CCO2 profiles. A phase
approximating steady-state production of CO2 (efflux) occurred between 14:30 and 18:30 h on Day 1,
and phases approximating steady-state consumption of CO2 (influx) occurred between 02:30 and
06:30 h on Day 1, and from 22:30 to 02:30 h on Day 2. During each of these steady-state phases,
changes in both CO2 concentration and δ13CCO2 profiles were minimal (Figure 4.4). Consequently,
during these periods, the rate of CO2 efflux or influx must equal the rate of CO2 production or
consumption, respectively. Furthermore, the δ13CΔCO2 of the net CO2 flux must reflect the δ13CCO2 of
the net CO2 being produced or consumed.
The phase of steady-state CO2 efflux (14:30 – 18:30 h on Day 1) corresponded to a period when soil
temperature at 5 cm depth remained relatively constant around its daily maximum, and soil
temperature at 15 cm depth increased at its maximum rate (Figure 4.3C and D). Similarly, the phase
of steady-state CO2 influx on Day 1 (between 02:30 and 06:30 h) corresponded to a period when soil
temperature at 5 cm depth remained relatively constant around its daily minimum, whilst soil
temperature at 15 cm depth decreased at its maximum rate. The approximate steady-state influx
92
period between 22:30 and 02:30 h on Day 2 had similar changes in soil temperature at 15 cm depth,
although soil temperature at 5 cm depth decreased during this period, rather than remaining
relatively constant.
During the period of steady-state efflux of CO2, the δ13CΔCO2 of the efflux, and hence the δ13CCO2 of CO2
being produced in the soil, averaged −3.8‰. During the two phases of steady-state influx, the δ13CCO2
of the CO2 consumed within the soil averaged −11.4‰. An additional value for the δ13CCO2 of CO2
produced can be obtained from the 14:30 h sampling on Day 2. The δ13CΔCO2 of the CO2 efflux at this
time is considered to reflect steady-state production of CO2, as despite not spanning a four-hour
period of steady-state due to day-to-day variation in soil temperature changes, the sampling at 14:30
h on Day 2 coincided with a period over which soil temperature at 5 cm depth remained relatively
constant around its daily maximum, whilst soil temperature at 15 cm depth increased at its
maximum rate (Figure 4.3C and D). These were the same conditions that prevailed during the period
of steady-state efflux on Day 1. The δ13CΔCO2 of the efflux at 14:30 h on Day 2 was −8.0‰.
Together with data from the steady-state efflux period on Day 1, these results suggest that during
phases of CO2 production, the δ13CCO2 of the net CO2 produced was approximately −5.2‰. In contrast,
the average δ13CCO2 of the net CO2 consumed during steady-state consumption phases was more
depleted (−11.4‰). These values are slightly lighter and heavier, respectively, than the average
δ13CΔCO2 values predicted from CO2 concentration and δ13CCO2 gradients during the steady-state
periods of CO2 efflux (−2.2‰) and influx (−15.8‰). Regardless, it appears that the net CO2 produced
during warm periods was isotopically heavier than the net CO2 consumed at cold times of the day.
This pattern cannot be accounted for by an isotopically light biological component.
If, during steady-state efflux (at times of warming), exsolution of relatively isotopically heavy CO2 was
accompanied by biological production of isotopically light CO2, the resulting flux would have an
isotopic composition intermediate between the two, i.e. lighter than that of the exsolved CO2, but
heavier than that of biologically-respired CO2. At times of cooling, if dissolution of CO2 and biological
respiration occurred simultaneously, the soil atmosphere would experience a loss of relatively heavy
CO2 and a gain of relatively light CO2. If this occurred during a period of steady-state influx, the net
δ13CCO2 of CO2 being absorbed would be highly enriched, and would have to be balanced by a highly
enriched surface influx. Therefore, the simultaneous occurrence of biological respiration with
abiotically-controlled CO2 exsolution or dissolution would, at steady-state, be evident as effluxes that
were isotopically lighter than influxes. This is the opposite of what was observed.
Instead of a biological explanation, the pattern of relatively heavy CO2 production (effluxes) and
relatively light CO2 consumption (influxes) can be explained by the kinetic fractionation that occurs as
a result of preferential exsolution of 13CO2 and preferential dissolution of 12CO2 in the CO2 (g) – CO2
93
(dissolved)
system. Under conditions of equilibrium fractionation, exsolution of CO2 concentrates 12CO2 in
the dissolved phase, making the δ13CCO2 of gaseous CO2 1.17‰ heavier than dissolved CO2 (H2CO3) at
1°C (Vogel et al., 1970). Therefore, during phases of exsolution, kinetically fractionated CO2 effluxes
would be isotopically heavier than the CO2 influxes during periods of dissolution, as was observed.
This phenomenon of exsolution of heavier CO2 and dissolution of lighter CO2 requires a long-term
shift in the isotopic composition of the dissolved inorganic C pool, as 12CO2 is being preferentially
stored in soil solution, whereas 13CO2 is being preferentially removed. As this fractionation effect is
induced by increases and decreases in soil temperature, driving exsolution and dissolution of CO2, it
may be the cause of the trend in δ13CCO2 profiles of increasing δ13CCO2 with depth (Figure 4.4). This
trend essentially favours 13CO2 diffusing out of the soil, and 12CO2 diffusing into the soil. That this
isotopic disequilibrium persists well into mid-summer is evidence of the ‘isotopic inertia’ associated
with the very large dissolved inorganic C pool at high soil pH (Gamnitzer et al., 2011), relative to the
size of the fluxes.
4.4.4 Data reliability
The surface CO2 flux rates measured in this study appear to be reliable measurements, as in addition
to their linear relationship with subsurface CO2 concentration gradients, artefacts associated with
closed chamber sampling of surface fluxes in comparatively higher-flux environments are not
apparent. The maximum change in headspace chamber CO2 concentration over the sampling period
was 46 µL L-1, and changes in δ13CCO2 were also low (up to 1.4‰). As such, any pressure changes
caused by CO2 concentration increases or decreases within chambers are likely to be negligible.
Furthermore, the small changes in CO2 concentration and δ13CCO2 during surface flux sampling mean
that chamber-to-soil feedbacks (caused by changes in headspace CO2 concentration and δ13CCO2
values), and associated errors caused by lateral diffusion (Nickerson and Risk, 2009a, 2009b), are
unlikely to have introduced errors in calculated surface CO2 flux rates and the δ13CCO2 of the net flux.
This is supported by the fact that calculated values of the δ13CCO2 of net CO2 fluxes closely matched
δ13CΔCO2 values calculated from CO2 concentration and δ13CCO2 gradients (Figure 4.3B). These gradients
were determined from subsurface sampling tubes located adjacent to but not beneath the
chambers.
4.4.5 Implications
The diel variability in surface CO2 flux rates has profound implications for interpretation of soil
biological activity and quantification of soil C turnover times. Aside from other studies investigating
diel soil CO2 fluxes (Parsons et al., 2004; Ball et al., 2009), previous measurements of soil CO2 fluxes
were often made around mid-day (Barrett et al., 2006c; Barrett et al., 2006b; Barrett et al., 2008a),
94
coinciding with the warmest part of the day, and were therefore assumed to represent the maximum
rate of soil respiration. Other studies either did not specify the time(s) of sampling (Burkins et al.,
2001; Gregorich et al., 2006; Ball et al., 2011) or specifically made measurements on days with
contrasting air temperatures in order to investigate the relationship between soil CO2 flux and
temperature (Elberling et al., 2006), but did not investigate this effect through a diel cycle. Many of
these studies found relationships between the rate of CO2 efflux and soil temperature. However,
given the data presented in this study, any CO2-temperature relationship cannot be used to solely
explain differences in rates of heterotrophic respiration. Whilst there may have been a biotic
component to the flux, CO2 effluxes in these studies will also have been influenced by abiotic
processes.
The potential for misinterpretation of surface CO2 fluxes is exemplified by the data from 10:30 h on
Day 1 (Figure 4.3). At this time, the soil was warming, the CO2 flux was positive, and had a δ13CΔCO2 of
−26.1‰. Without the insight into diel variability and dynamic fractionation effects demonstrated in
this study, these values would be incorrectly interpreted as a flux of biological origin.
The dynamic nature of CO2 production and consumption in dry, low C soils such as those at Site B,
and associated changes in δ13CΔCO2, mean that periods approximating steady-state, and therefore
periods during which the surface CO2 flux and its isotopic composition are a true representation of
the soil CO2 being produced, are short-lived. Consequently, studies using stable C isotopes to
quantify heterotrophic respiration would require at least semi-continuous monitoring to identify
periods of steady-state. This approach requires that both the isotopic composition of the source
organic material and the abiotically-produced CO2 are known. The latter is as yet unknown, and must
be the focus of further research. However, as the isotopic composition of CO2 that is dissolved into
and exsolved from soil solution will be influenced by kinetic fractionation, it is likely to be dependent
on the rate of soil warming and cooling, thus adding a further requirement of detailed soil
temperature measurements during sampling periods.
Alternatively, semi-continuous monitoring of sufficient frequency and precision could be used to
calculate net CO2 fluxes over diel cycles, and any net positive flux could be attributed to
heterotrophic respiration. However, this assumption relies on any additional sources of CO2 such as a
flux of geological origin (Risk et al., 2013), or an enhanced flux due to wind pumping being excluded.
It also requires that there is no net change in soil inorganic C storage as a result of changing soil
conditions (primarily temperature, moisture content and pH).
95
4.5 Summary and conclusions
Surface CO2 fluxes from the dry, low C, high pH, low EC soils at Site B were low, and varied diurnally
between −0.04 and 0.06 µmol m−2 s−1. The diel pattern of positive and negative surface CO2 fluxes
was such that the net flux over six overlapping 24-h periods was not significantly different from zero.
Diel variation in soil temperatures was the primary driver of surface CO2 fluxes, which were
predominantly diffusion-driven in response to CO2 concentration gradients generated by dissolution
and exsolution of CO2 in accordance with Henry’s Law.
The isotopic composition of surface CO2 fluxes was extremely variable, ranging between −2.4 and
−31.8‰. This was due to dynamic fractionation effects arising as a result of the interaction between
different diffusion coefficients for 12CO2 and 13CO2 and transient CO2 concentration gradients. As
such, dynamic fractionation, rather than heterotrophic respiration, accounted for the highly depleted
positive surface CO2 fluxes which occurred during warm parts of the day. At these times, transient
CO2 concentration and δ13CCO2 gradients meant that neither the flux rate nor the isotopic
composition of the surface CO2 flux was representative of soil CO2 production at that time.
Steady-state phases of CO2 production and consumption, during which the isotopic composition of
the surface CO2 flux reflected that of the CO2 being produced or consumed within the soil, were
short-lived. These phases were limited to periods of efflux and influx during which soil temperature
at 15 cm depth increased or decreased at its maximum rate, respectively. Phases of steady-state
showed that soil CO2 effluxes were isotopically heavier than CO2 influxes. This pattern of behaviour is
consistent with kinetic fractionation associated with Henry’s Law-controlled exsolution and
dissolution of CO2 into soil solution, but inconsistent with simultaneous production of an isotopically
light biological component to the flux.
All lines of evidence (no net flux over diel periods, dynamically fractionated depleted surface fluxes,
and the relatively heavy isotopic composition of CO2 produced) suggest that surface CO2 fluxes were
driven by abiotic mechanisms, and there was no readily distinguishable biological component to the
surface CO2 flux at this site. Consequently, previous measurements of surface CO2 fluxes from Dry
Valley soils with similar characteristics are likely to have interpreted the abiotic surface CO2 flux as a
biological flux, thus overestimating the rate of heterotrophic respiration.
In addition, the relative magnitude of surface fluxes and subsurface storage fluxes shown in this
chapter indicate that the closed-system assumption invoked in Chapter 3 is too strict a simplification.
This violation probably accounts for the discrepancies between measured and modelled subsurface
CO2 concentrations in Chapter 3, particularly at shallower depths (Chapter 3, Figure 3.12).
Furthermore, estimates of biotic and abiotic contributions to subsurface CO2 at the moist, relatively
96
high organic C site (Site A) were based on a simple isotope mixing model incorporating two internal
sources: a biotic component derived from metabolisation of soil organic C, and an abiotic source with
a range of isotopic compositions under various scenarios (Chapter 3, Table 3.3). This chapter has
shown that the direction of surface CO2 fluxes makes an important contribution to changes in
subsurface soil CO2 concentrations, and dynamic fractionation of these CO2 fluxes is the major
determinant of the changes in the isotopic composition of soil CO2. As such, estimates of the
biological contribution to subsurface soil CO2 production at Site A made in Chapter 3 are likely to be
in error.
The generality of these conclusions, along with further discussion of diel variability in surface CO2
fluxes and implications for soil respiration measurements in the McMurdo Dry Valleys is considered
in Chapter 6. This chapter follows an additional results chapter which presents new data from an
equally intensive sampling regime for Site A, a site with higher levels of organic C and soil moisture
but with similar soil chemistry to that at Site B.
97
98
Chapter 5
Quantifying diel variations in biotic and abiotic soil CO2 fluxes,
Taylor Valley, Antarctica (II): The influence of a contemporary
organic carbon source
5.1 Introduction
Chapter 4 showed that diel variability in surface CO2 fluxes at a site in Taylor Valley with no
contemporary lacustrine organic C source was predominantly driven by abiotic processes. Surface
CO2 fluxes were primarily diffusion-driven in response to CO2 gradients generated by dissolution and
exsolution of CO2 in accordance with Henry’s Law. As such, diel variation in soil temperatures was the
primary driver of variation in surface CO2 flux rates. Variation in subsurface CO2 concentration and
δ13CCO2 values were influenced by surface CO2 fluxes, and vice versa. Dynamic fractionation, which
determined the isotopic composition of the surface CO2 flux, occurred during subsurface diffusion of
CO2, with feedback from the direction and isotopic composition of the surface flux acting to change
the isotopic composition of soil CO2. Despite several occurrences of highly depleted surface CO2
fluxes (as low as −31.8‰), no distinguishable biological component to surface CO2 fluxes was
detected. The highly depleted surface fluxes, rather than being driven by steady state production of
highly depleted biologically-produced CO2, were a result of dynamic fractionation of a relatively
enriched CO2 source, and occurred when CO2 concentration and δ13CCO2 gradients were of opposite
sign. Further support for an abiotic driver of CO2 fluxes was seen in the δ13CΔCO2 of surface fluxes
during periods of steady state CO2 production. The relatively enriched δ13CΔCO2 (−5.2‰) of surface
fluxes at these times is consistent with the kinetic fractionation that occurs as a result of preferential
exsolution of 13CO2 in the CO2 (g) − CO2 (dissolved) system.
This chapter aims to further advance the understanding of Dry Valley soil CO2 dynamics by applying
knowledge gained from Site B, a site with no contemporary lacustrine organic C source, to Site A,
located adjacent to a small lake in Taylor Valley. Algae formed in the lake during the austral summer
are driven to the shore by wind and wave action, and subsequently blown onto nearby soils. This
contemporary organic C source provides an opportunity to investigate CO2 fluxes in soils with above
average organic C, and thereby identify the relative contributions of biotic and abiotic processes to
surface CO2 fluxes at such sites. As in Chapter 4 (Site B), four-hourly measurements of surface CO2
flux rates and subsurface CO2 concentration and δ13CCO2 profiles were made over a 48-h period. This
chapter is focused on Site A, the same “Site A” in Taylor Valley as that studied in Chapter 3.
99
The field work which generated the data considered in this chapter was undertaken concurrently
(during the 2009/10 austral summer) with the field work providing the data considered in the
previous chapter. This work was conducted a full year after the field work which yielded results for
Chapter 3. The materials and methods utilised were the same as those carried out at Site B during
the 2009/10 summer (Chapter 4), although differences pertinent to Site A are noted below.
5.2 Materials and methods
5.2.1 Study area
At the time of gas sampling, soils at Site A were ice-cemented below 35 cm depth, which was the
same as the depth of ice cement measured during the sampling period in 2008/09.
5.2.2 Environmental variables
During the intensive 48-h gas sampling period, all data were recorded and logged every 2 minutes by
a CR1000 data logger (Campbell Scientific Inc., Logan, Utah, USA).
Soil temperature and moisture content measurements
Soil temperature and volumetric moisture content at 5, 15 and 25 cm depth were measured using
Hydra Probes (Hydra Probe II, Stevens Water Monitoring Systems Inc., Portland, Oregon, USA). As in
the 2008/09 sampling period at Site A, an intermittent fault occurred in the probe installed at 5 cm
depth. Consequently, the soil temperature and moisture content data shown for 5 cm depth are
from an additional data logger installed in an adjacent polygon. The additional logger included the
same equipment, with Hydra Probes installed at the same soil depths, and was located ~ 10 m SSW
of the primary logger at Site A. The soil and environmental conditions in the two polygons were very
similar. Consequently, data from the probe at 5 cm depth in the secondary polygon are considered to
be an appropriate surrogate for the missing data.
The rate of change in soil temperature (ΔT/Δtt) was calculated as the difference in temperatures (T)
measured ten minutes apart, as follows:
Air temperature
∆𝑇𝑇 𝑇𝑇𝑡𝑡+5 𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑇𝑇𝑡𝑡−5 𝑚𝑚𝑚𝑚𝑚𝑚
=
.
∆𝑡𝑡𝑡𝑡
10 𝑚𝑚𝑚𝑚𝑚𝑚
(5.1)
Surface air temperature was measured 2 mm above the soil surface (CS215-L, Campbell Scientific
Inc., Logan, Utah, USA).
100
Wind speed
Wind speed (#40C Anemometer, NRG Systems, Hinesburg, Vermont, USA) and direction (W200P
Potentiometer Windvane, Vector Instruments, North Wales, UK) were measured 1.2 m and 1.3 m
above the soil surface, respectively.
5.2.3 Soil CO2 sampling and analysis
The experimental set-up and sampling for surface soil CO2 fluxes proceeded as at Site B in 2009/10
(Chapter 4). Three chambers were installed in each of two adjacent polygons at Site A; four 20 ml gas
samples were taken from each chamber at 15 min intervals over a 45 min period. The polygons in
which gas samples were taken were located ~ 45 m south-west of the sampling site utilised at Site A
in 2008/09, on the same geomorphic surface.
Soil profile CO2 sampling also proceeded as at Site B, with subsurface sampling tubes installed
adjacent to each of the three chambers in each polygon at depths of 5, 10, 15, 20 and 30 cm.
Additionally, one sampling tube was installed horizontally, 1 cm above the desert pavement, in each
polygon. Gas samples from these tubes provided an atmospheric CO2 sample from just above the soil
surface.
Soil CO2 sampling began > 70 h after installation of equipment. Gas samples were taken every four
hours over a 48-h period, starting at 14:30 h on December 29th, 2009. Gas sampling at Site A
proceeded concurrently with gas sampling at Site B, thus data from each sampling time are directly
comparable between the two sites. However, localised effects of topographic shading dictate that
the timing of maxima and minima in air and soil temperatures were slightly offset. A selection of
samples were analysed for CO2 concentration and δ13CCO2 using a Gas-bench II connected to a
DeltaPlus Advantage isotope ratio mass spectrometer (both Thermo Finnigan, Bremen, Germany) at
the Macaulay Land Use Research Institute in Aberdeen, Scotland, within 5 months of being collected.
Budget constraints meant that it was not possible to analyse all of the samples collected. Standard
reference gas samples collected in the field and standard reference gases prepared in New Zealand
and included with samples sent to Scotland showed no sign of storage or transport effects. The
analytical precision for CO2 concentration and δ13CCO2 values was ≤5 µL L-1 and ≤0.05‰, respectively.
5.2.4 Soil sampling, physical and chemical analyses
The gas samples at Site A were taken within 30 m of the soil pit sampled at Site A in 2008/09. Based
on the proximity of the 2009/10 gas sampling site to the previously sampled and analysed soil pit, the
uniformity of the geomorphic surface on which samples were taken, and the similar physical
appearance of soil pits dug and described in polygons adjacent to the 2009/10 gas sampling site
101
(Figure 5.1), the soil physical and chemical characteristics at Site A were considered to be accurately
represented by the data presented in Chapter 3.
Figure 5.1
Morphological comparison between soil pits adjacent to gas sampling sites at Site A in
the 2008/09 and 2009/10 austral summers. At the time of sampling, both soils (Typic
Haploturbels) were ice-cemented below 35 cm depth. The moistening of the upper ~ 2
cm of the 2009/10 profile is from melting of recent snowfall.
5.2.5 Statistical analyses and calculations
Surface CO2 fluxes
As for Site B (Chapter 4), surface CO2 fluxes were determined from changes in headspace CO2
concentrations during the 45 min sampling period as follows:
𝑛𝑛𝐶𝐶𝑂𝑂2 = [𝐶𝐶𝑂𝑂2 ]
𝑃𝑃𝑃𝑃
𝑅𝑅𝑅𝑅𝑅𝑅
(5.2)
where nCO2 is the number of moles of CO2 emitted per m2 (µmol m-2), [CO2] is headspace CO2
concentration (µL L-1), P is atmospheric pressure (101325 Pa), V is the chamber volume (m3), R is the
ideal gas constant (8.314 J K-1 mol-1), T is air temperature (K), and A is the area covered by the
chamber (0.0177 m2).
Surface CO2 flux rates (µmol m-2 s-1) were calculated by linear regression (Microsoft Excel® 2010) of
surface CO2 fluxes (µmol m-2) against time over which samples were collected (s). Surface flux rates
102
that were significantly different to zero (p < 0.1) were subsequently used to calculate the net CO2
gained or lost from the surface chambers (Δ[CO2]), and the average δ13C of the net CO2 gained or lost
(δ13CΔCO2). Each significant surface CO2 flux, which may be positive or negative, was considered
individually, and calculations were as follows:
∆[𝐶𝐶𝑂𝑂2 ] = [𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖
(5.3)
where [CO2]final and [CO2]initial are the final and initial CO2 concentrations (µL L-1) measured from the
final and initial CO2 samples taken from the chamber, respectively.
Using a mass balance approach,
∆[𝐶𝐶𝑂𝑂2 ] ∙ 𝛿𝛿 13 𝐶𝐶∆𝐶𝐶𝐶𝐶2 = [𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ,
(5.4)
where δ13CΔCO2 (‰) is the δ13C of the net CO2 gained or lost between the final and initial
measurements, and δ13C[CO2]final and δ13C[CO2]initial are the δ13CCO2 values (‰) measured from the final
and initial CO2 samples taken from the chamber, respectively.
Therefore,
𝛿𝛿 13 𝐶𝐶∆𝐶𝐶𝐶𝐶2 =
[𝐶𝐶𝑂𝑂2 ]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − [𝐶𝐶𝑂𝑂2 ]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝛿𝛿 13 𝐶𝐶[𝐶𝐶𝐶𝐶2]𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖
∙
∆[𝐶𝐶𝑂𝑂2 ]
(5.5)
Subsurface storage CO2 fluxes
Subsurface storage CO2 fluxes were estimated by calculating differences in the average soil profile
CO2 concentration between consecutive sampling times. The average soil profile CO2 concentration
was determined by using an analogue integration technique whereby the area under a paper plot of
CO2 concentration versus depth at each sampling time was weighed to determine the integral
��������
[𝐶𝐶𝑂𝑂2 ] =
30
∫0 [𝐶𝐶𝑂𝑂2 ] 𝑑𝑑𝑑𝑑
30
(5.6)
��������
where z is depth in cm and [𝐶𝐶𝑂𝑂
2 ] is the profile average CO2 concentration between 0 and 30 cm
��������
depth. The difference in [𝐶𝐶𝑂𝑂
2 ] between consecutive sampling times was used to determine the
number of moles of CO2 (nCO2; µmol m-2) gained or lost over the period between sampling times using
the ideal gas law:
𝑛𝑛𝐶𝐶𝑂𝑂2 = ∆[𝐶𝐶𝑂𝑂2 ]
𝑃𝑃𝑃𝑃
𝑅𝑅𝑅𝑅
(5.7)
103
where Δ[CO2] is the difference in the profile average CO2 concentration between sampling times (µL
L-1), P is atmospheric pressure (101325 Pa), V is the soil gas volume (m3 m-2; calculated by multiplying
soil depth (0.3 m) by the air-filled porosity (ε = 0.4)), R is the ideal gas constant (8.314 J K-1 mol-1) and
T is soil temperature (K). This was converted to a flux rate (µmol m-2 s-1) by dividing by the time (s)
over which nCO2 was gained or lost.
Calculated surface CO2 fluxes and their C isotopic composition
As a test of surface CO2 flux rates calculated from surface chamber CO2 measurements, surface CO2
flux rates (F) were also estimated using near-surface CO2 concentration gradients to solve Fick’s Law
𝐹𝐹 = −𝐷𝐷𝑒𝑒
𝑑𝑑𝑑𝑑
,
𝑑𝑑𝑑𝑑
(5.8)
where De is the effective diffusivity of CO2 (cm2 s-1), and dC/dz is the CO2 concentration gradient. The
value of De used (0.032 cm2 s-1) was derived from the relationship between surface CO2 flux rates and
near-surface CO2 concentration gradients at Site B, and it is considered that this value can be reliably
extrapolated to Site A as the soils were similarly coarse textured, with low levels of moisture.
In a diffusion-driven system, the δ13CΔCO2 of surface CO2 fluxes is determined by subsurface CO2
concentration and δ13CCO2 gradients. Using data from subsurface profiles, the δ13CΔCO2 of surface CO2
fluxes can therefore be determined as follows:
𝑅𝑅13 =
𝐹𝐹13
𝐹𝐹12
(5.9)
where R13 is the ratio of 13C to 12C in the CO2 flux, and F13 and F12 refer to the flux of 13CO2 and 12CO2,
respectively. The full derivation of this calculation is given in Chapter 4 (equations 4.8 – 4.26), which
shows that:
𝑅𝑅13
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑅𝑅𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑑𝑑(𝛿𝛿 13 𝐶𝐶)
[𝐶𝐶𝑂𝑂2 ]
𝐽𝐽(1 + 𝐽𝐽) + 1000
𝑑𝑑𝑑𝑑
= 0.9956 � 𝑑𝑑𝑑𝑑
�,
𝑑𝑑(𝛿𝛿 13 𝐶𝐶)
𝑑𝑑[𝐶𝐶𝑂𝑂2 ]
𝑅𝑅
[𝐶𝐶𝑂𝑂2 ]
(1 + 𝐽𝐽) − 𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
1000
𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑
(5.10)
where J = (δ13CCO2/1000 + 1) RVPDB.
104
Mixing model
As in Chapter 3, a mixing model (Fry, 2006) was used to constrain the biological contribution (fbiotic) to
surface CO2 fluxes as follows:
𝑓𝑓𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 =
𝛿𝛿 13 𝐶𝐶∆𝐶𝐶𝐶𝐶2 − 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
𝛿𝛿 13 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 − 𝛿𝛿 13 𝐶𝐶𝐶𝐶𝐶𝐶2−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
(5.11)
where δ13CΔCO2 is the isotopic composition of the net surface CO2 flux (‰), δ13Cbiotic is the
concentration-weighted δ13C of soil organic C (between 0 and 30 cm depth) at Site A (−22.8‰;
Chapter 3, Figure 3.8), and δ13CCO2-abiotic is the average δ13CΔCO2 of the surface CO2 efflux during periods
of steady state at Site B (−5.2‰; Chapter 4, section 4.4.3). As soil CO2 dynamics at Site B appear to be
abiotically-driven, this value is considered to represent abiotic production of CO2 within the soil.
5.3 Results
5.3.1 Environmental variables
At the time of sampling (late December 2009), soils at Site A were ice-cemented below 35 cm. Air
and soil temperatures varied diurnally, with the latter exhibiting progressively greater lags and
damping in soil temperature response to surface heating and cooling with increasing depth in the
profile (Figure 5.2A). Air and soil temperatures were slightly warmer at Site A than Site B, with air
temperature at Site A ranging from −3.6 to 8.7°C, and soil temperatures at 5 cm, 15 cm and 25 cm
ranging from −0.8 to 11.0°C, 1.0 to 5.3°C and 0.8 to 2.9°C, respectively.
105
Figure 5.2
Soil temperature (A) and volumetric moisture content (B) during the 48-h period over
which gas samples were taken, Site A, Taylor Valley. Data plotted for moisture content
are 30 min. moving averages. The narrow vertical grey boxes denote the periods during
which gas samples were taken.
106
Soil moisture content remained relatively constant throughout the sampling period (Figure 5.2B), and
increased with depth in the profile, averaging 0.8, 4.6 and 7.3% at 5, 15 and 25 cm, respectively. This
was significantly higher than soil moisture content at Site B, which averaged 0.8% over the three
depths at which measurements were made. Qualitative observations of limited melting of snow
patches and lake ice in late December 2009 suggested that the summer was cooler than the previous
summer of 2008/09. This was corroborated by the lack of ponded surface water evident at sites close
to Site A where it had been observed the previous summer, and low soil moisture contents measured
at Site B. However, soil moisture contents at Site A were similar to those measured during the
2008/09 sampling period.
5.3.2 CO2 dynamics
Surface CO2 fluxes
Surface CO2 flux rates at Site A were highly spatially and temporally variable, ranging from −0.44 to
0.17 µmol m-2 s-1 (Table 5.1). However, only 25% of the surface CO2 fluxes calculated were
statistically significant, and, of these, 83% occurred at either the 06:30 or 10:30 h sampling periods.
At 14:30, 18:30, 22:30 and 02:30 h, surface chamber CO2 concentrations generally varied nonsystematically throughout the 45-min sampling period (Figure 5.3), hence surface flux rates were
unable to be calculated. Consequently, it is difficult to determine the timing of maximum and
minimum surface CO2 fluxes and their relationship with wind speed, soil temperature, and the rate of
change in soil temperature (Figure 5.4).
In the absence of a large proportion of significant linear relationships between chamber headspace
CO2 concentration and time, surface CO2 fluxes were predicted according to Fick’s Law (5.8). At times
of low flux (06:30 and 10:30 h), the measured and predicted fluxes were in good agreement (Figure
5.4A). As such, it appears that the relationship between subsurface CO2 concentration gradient and
surface CO2 flux rate (and hence De) derived from Site B is transferable to Site A. However, the
statistically significant high-flux measurements at 18:30 and 22:30 h on Day 1 were lower than the
predicted fluxes at those times, with the discrepancy at 22:30 h being such that the measured and
predicted fluxes were of opposite sign. Furthermore, these two occurrences of statistically significant
high-flux measurements originated from a single chamber. As the chamber that produced the
significant results differed at each of the two sampling times, it suggests that the apparently
significant fluxes may result from a coincidental artefact, rather than representing a statistically
significant, repeatable result.
107
Table 5.1
Date
29/12/10
Surface soil CO2 fluxes at Site A, Taylor Valley.
Day 1
Sampling Replicate
time (h)
14:30
a
b
c
d
CO2 flux
Date
0.034
0.042
0.005
0.047
30/12/10
(µmol m-2 s-1)
Day 2
Sampling Replicate
time (h)
14:30
a
b
c
d
CO2 flux
(µmol m-2 s-1)
−0.018
−0.168
0.013
−0.080
29/12/10
18:30
a
b
c
d
0.133**
0.033
−0.026
−0.001
30/12/10
18:30
a
b
c
d
−0.017
0.033
−0.001
−0.001
29/12/10
22:30
a
b
c
d
0.031
−0.008
−0.020
−0.197**
30/12/10
22:30
a
b
c
d
−0.012
−0.018
0.017
−0.077
30/12/10
02:30
a
b
c
d
−0.071
0.102
0.010
0.173
31/12/10
02:30
a
b
c
d
−0.012
−0.439
−0.114
−0.227
30/12/10
06:30
a
b
c
d
−0.017
−0.031**
−0.057*
−0.026
31/12/10
06:30
a
b
c
d
−0.018
0.065*
−0.020
−0.030*
30/12/10
10:30
a
0.027
b
0.040**
c
0.022
d
0.040**
* represents P < 0.1; ** represents P < 0.05.
31/12/10
10:30
a
b
c
d
0.032**
0.031*
0.033**
0.060**
The average δ13CCO2 of the four gas samples taken from surface chambers during the 45-min sampling
periods ranged from −8.6‰ to −11.4‰. Where surface fluxes were statistically significant, changes in
δ13CCO2 between samples taken at times 0 and 45 min were used to calculate the δ13CΔCO2 of the
surface flux (5.5). The δ13CΔCO2 of significant surface CO2 fluxes ranged from −9.3 to −29.8‰ (Figure
5.4B). The predicted δ13CΔCO2 of surface fluxes, based on CO2 concentration and δ13CCO2 gradients
(5.10), ranged from −8.1 to −25.3‰. However, the lack of significant surface CO2 fluxes limited
comparisons between measured and predicted δ13CΔCO2 values. On several occasions, the measured
and predicted δ13CΔCO2 values were in good agreement, whereas at other times, they were
considerably different.
108
Figure 5.3
Selection of CO2 time-series data from surface headspace chambers, Site A, Taylor
Valley, showing variation in CO2 concentration over the 45-min sampling period. Note
that data from each sampling period are plotted on different CO2 concentration scales.
109
Figure 5.4
(A) Average significant measured surface CO2 flux rates, predicted surface CO2 flux rates,
and subsurface soil CO2 storage flux rates at Site A, Taylor Valley. Error bars represent
the standard error of the mean. Subsurface soil CO2 storage fluxes are shown as grey
bars, which represent the net CO2 flux over the 4-h period between sampling. (B)
Isotopic composition (δ13CΔCO2) of measured and predicted surface CO2 fluxes at Site A,
Taylor Valley. Predicted values are calculated based on subsurface CO2 concentration
110
and δ13CCO2 gradients at the time of sampling. Error bars represent the standard error of
the mean. The dashed horizontal line represents the average δ13CCO2 of ambient
atmospheric CO2 (−8.8‰). (C) Soil temperature at 5 and 15 cm depth (left hand axis) and
wind speed (right hand axis) during the gas sampling period, Site A, Taylor Valley. (D)
Rate of change in soil temperature at 5 and 15 cm depth during the gas sampling period,
Site A, Taylor Valley. Data plotted are 30 min moving averages.
The dashed vertical lines correspond to the four-hourly intervals at which gas sampling
began.
Subsurface CO2 profiles and fluxes
Values for CO2 concentration and δ13CCO2 at 0 cm depth varied throughout the 48-h sampling period
(Figure 5.5), suggesting that the 0 cm sampling tube (1 cm above the soil surface) was located within
a partially mixed surface boundary layer. The highest CO2 concentrations and most depleted δ13CCO2
values occurred at 22:30 and 02:30 h on both days.
Temporal changes in CO2 concentration and δ13CCO2 profiles throughout the two 24-h sampling
periods at Site A were highly variable (Figure 5.5). Between 06:30 and 14:30 h, CO2 concentration
profiles were generally symmetrical about a vertical line corresponding to the local ambient
atmospheric CO2 concentration (400 ± 4 µL L-1, as revealed by data from 0 cm sampling tubes at Site
B). CO2 concentrations were below that of atmospheric CO2 concentration throughout the profile at
06:30 h, and in the lower part of the profile at 10:30 h, after which they increased to be above that of
atmospheric CO2 concentration throughout the profile by 14:30 h. Between 14:30 and 02:30 h, soil
CO2 concentrations throughout the profile remained greater than that of ambient atmospheric CO2
concentration. The highest subsurface CO2 concentrations were measured at 02:30 h on Day 2,
reaching a maximum of 700 µL L-1 at 15 cm depth. The greatest change in subsurface CO2
concentrations occurred between 02:30 and 06:30 h, when concentrations decreased from
maximum to minimum values (below atmospheric CO2 concentration).
111
Figure 5.5
Soil profile CO2 concentration and δ13CCO2 at Site A, Taylor Valley. (A) and (B) represent
data from Day 1 and Day 2, respectively. Data shown represent averages from each
sampling time (n = 1 or 2 for 0 cm samples. At all other depths, n = 4). Error bars
represent the standard error of the mean.
112
Depth profiles of δ13CCO2 were also highly variable (Figure 5.5). Between 06:30 and 14:30 h, δ13CCO2
values remained close to that of ambient atmospheric CO2 (−8.8 ± 0.2‰) throughout the profile,
although δ13CCO2 values tended to increase with depth. Between 14:30 and 02:30 h, δ13CCO2 became
significantly more depleted than that measured between 06:30 and 14:30 h. The most depleted
δ13CCO2 values (as low as −13.7‰) were measured at 02:30 h, although the uncertainty associated
with δ13CCO2 values on Day 2 indicates that there was high spatial variability in δ13CCO2 values at this
time. As with subsurface CO2 concentrations, the greatest change in subsurface δ13CCO2 values
occurred between 02:30 and 06:30 h, when δ13CCO2 shifted from its most depleted to close to that of
ambient atmospheric CO2. Changes in δ13CCO2 values coincided with changes in CO2 concentrations,
with δ13CCO2 generally becoming more depleted in 13C as CO2 concentrations increased, and more
enriched in 13C as CO2 concentrations decreased. This is the opposite of what was observed at Site B.
Subsurface storage CO2 fluxes were calculated based on the change in the profile average CO2
concentration between sampling times (5.7), and ranged between −0.07 and 0.05 µmol m-2 s-1
(Figure 5.4A). The only negative subsurface storage fluxes occurred during the period between 02:30
and 06:30 h each day.
5.3.3 Linking surface and subsurface CO2 dynamics
Day 1 (Figure 5.4 and Figure 5.5A)
14:30 – 18:30 h
Between 14:30 and 18:30 h, the subsoil gained CO2. At the start of the interval, subsurface CO2
concentrations were above ambient atmospheric CO2 concentration; concentration increased over
the interval at all depths, with the most significant increases in the upper 10 cm of the profile. The
greatest increase was at 5 cm depth, with CO2 concentration increasing from 420 to 490 µL L-1.
Subsurface δ13CCO2 values became more depleted, with the largest shifts corresponding to the
concentration increases in the upper 10 cm of the profile. The maximum depletion in δ13CCO2
occurred at 5 cm depth, shifting from −8.7 to −10.7‰.
No significant surface CO2 fluxes were measured at 14:30 h, but at 18:30 h, there was one significant
surface flux at a rate of 0.13 µmol m-2 s-1. Surface CO2 fluxes calculated from subsurface CO2
concentration gradients increased from 0.07 µmol m-2 s-1 at 14:30 h to 0.20 µmol m-2 s-1 at 18:30 h,
meaning that the predicted flux at 18:30 h was around 30% higher than the single significant
measured flux.
The shift in soil profile δ13CCO2 to more depleted values meant that the subsurface δ13CCO2 gradient
switched from being positive at 14:30 h to negative at 18:30 h. Combined with the persistent positive
CO2 concentration gradient, the switch in δ13CCO2 gradient acted to drive a relatively depleted flux at
113
18:30 h, with the δ13CΔCO2 of the flux calculated from CO2 and δ13CCO2 gradients shifting from −11.2‰
at 14:30 h to −22.3‰ at 18:30 h. The δ13CΔCO2 of the predicted flux at 18:30 h was very similar to the
δ13CΔCO2 of the single measured significant flux (−20.1‰). Over this period, soil temperature at 5 cm
depth remained relatively constant (around 9°C), and soil temperature at 15 cm depth increased to
close to its daily maximum by the end of the interval.
18:30 – 22:30 h
Over this interval, the soil gained a small amount of CO2. Subsurface CO2 concentration increased at 5
cm depth to a daily maximum of 570 µL L-1 at 22:30 h, although this mean value has a high standard
error due to high spatial variability amongst samples. CO2 concentration decreased at 10 cm depth,
and did not change between 15 and 30 cm depth, thus CO2 concentrations throughout the profile
remained greater than that of atmospheric CO2. Changes in subsurface δ13CCO2 values corresponded
to changes in CO2 concentration; δ13CCO2 at 5 cm depth reached the daily minimum value of −12.0‰
at 22:30 h. However, as with CO2 concentrations, the δ13CCO2 at this depth was highly spatially
variable. δ13CCO2 became less depleted at 10 cm depth, and remained similar between 15 and 30 cm
depth.
The sign of the single measured significant surface CO2 flux shifted from positive at 18:30 h to
negative at 22:30 h (0.13 to −0.20 µmol m-2 s-1). This was in contrast to the predicted surface CO2 flux
rates, which were positive throughout this 4-h period, and reached a daily maximum at 22:30 h.
However, the high flux predicted at 22:30 h should be interpreted with caution, as there was large
uncertainty in the CO2 concentration value at 5 cm depth, from which the CO2 concentration gradient
used in the flux calculation was derived.
The δ13CΔCO2 of the significant measured surface fluxes (n = 1 at each sampling time) shifted from
−20.1‰ at 18:30 h to −29.8‰ at 22:30 h. The highly depleted flux measured at 22:30 h was
consistent with the relatively light efflux expected from the persistent combination of a positive CO2
concentration gradient and a negative δ13CCO2 gradient, although it was considerably more depleted
than that calculated from CO2 concentration and δ13CCO2 gradients at that time (−21.3‰). The
increase in surface CO2 flux rates over this period was coincident with rapidly decreasing soil
temperature at 5 cm depth, at times at close to its maximum negative rate of change. Soil
temperature at 15 cm depth remained close to its daily maximum throughout the interval.
22:30 – 02:30 h
Between 22:30 and 02:30 h, there was no net change in subsurface CO2 concentrations. Subsurface
CO2 concentration decreased at 5 cm depth, and increased between 10 and 30 cm depth, giving a
profile average CO2 concentration of 460 µL L-1, the same as that at 22:30 h. Subsurface δ13CCO2
114
became more enriched at 5 cm depth, but was significantly more depleted throughout the rest of the
profile. The maximum depletion occurred at 10 cm depth, with δ13CCO2 shifting from −9.5 to −11.8‰.
No significant surface CO2 fluxes were measured at 02:30 h, hence the δ13CΔCO2 of the net flux could
not be calculated from surface chamber data. Predicted surface CO2 flux rates decreased from 0.33
µmol m-2 s-1 at 22:30 h to 0.10 µmol m-2 s-1 at 02:30 h in response to the reduced concentration
gradient. The positive CO2 concentration and negative δ13CCO2 gradients persisted throughout the
interval, and the δ13CΔCO2 of the predicted surface flux at 02:30 h was highly depleted (−27.3‰). The
rate of change in soil temperatures at 5 and 15 cm depth were both negative, with the latter at its
maximum negative rate by the end of the interval.
02:30 – 06:30 h
During this period, the subsoil lost CO2 at a rate of −0.03 µmol m-2 s-1; the associated changes in
subsurface CO2 concentrations and δ13CCO2 were the most significant observed on Day 1. Subsurface
CO2 concentrations decreased at all depths, declining from a profile average of 460 µL L-1 at 02:30 h
to a profile average of 375 µL L-1 (below atmospheric CO2 concentration) at 06:30 h. δ13CCO2 became
more enriched at all depths, with the greatest shift (from −11.8 to −9.0‰) coinciding with the
greatest decrease in CO2 concentration at 10 cm depth. At 06:30 h, two significant surface CO2 fluxes
were measured, averaging –0.04 µmol m-2 s-1, whereas the surface flux rate predicted at this time
was −0.08 µmol m-2 s-1. The significant decrease in CO2 concentration between 02:30 h and 06:30 h
changed the concentration gradient from positive to negative, and the associated enrichment in
δ13CCO2 changed the δ13CCO2 gradient from negative to positive, thus driving a relatively depleted
predicted influx of CO2 (−25.4‰). This was in contrast to the relatively enriched average δ13CΔCO2 of
the measured surface influxes at 06:30 h (−9.3‰). Soil temperature at 5 cm depth remained close to
its daily minimum (1.1°C) throughout the interval, while soil temperature at 15 cm depth continued
to decrease at its maximum (negative) rate.
06:30 – 10:30 h
Between 06:30 and 10:30 h, the soil gained CO2, with subsurface CO2 concentrations increasing
between 0 and 20 cm depth. However, subsurface CO2 concentrations remained below that of
atmospheric CO2 beneath 10 cm depth. Subsurface δ13CCO2 values became slightly more enriched
between 5 and 30 cm depth. The profile average δ13CCO2 of −8.4‰ at 10:30 h was the most enriched
measured on Day 1.
Surface CO2 fluxes shifted from negative to positive over this period, with significant fluxes of 0.04
µmol m-2 s-1 measured from two chambers at 10:30 h. This was consistent with predicted surface
fluxes, which also shifted from negative (−0.08 µmol m-2 s-1) to positive (0.07 µmol m-2 s-1). The
115
change in surface flux rate was also accompanied by a significant shift in the δ13CΔCO2 of the surface
flux. At 06:30 h, the average δ13CΔCO2 of the measured CO2 influx was −9.3‰, which was substantially
more enriched than that predicted from CO2 concentration and δ13CCO2 gradients at that time
(−25.4‰). By 10:30 h, surface CO2 fluxes were positive, and the average δ13CΔCO2 of the measured CO2
efflux was −20.8‰, similar to the predicted δ13CΔCO2 value of −17.4‰. Soil temperature at 5 cm depth
increased from close to its daily minimum, and soil temperature at 15 cm depth decreased at a low
rate, reaching its daily minimum in the second half of the interval.
10:30 – 14:30 h
This interval completes a full 24-h cycle, thus data from the subsurface profile measured at 14:30 h
are shown on the “Day 2” profile (Figure 4.4B). Accumulation of CO2 in the subsoil continued at a
greater rate than during the previous 4-h interval, and by 14:30 h, subsurface CO2 concentrations
were greater than atmospheric CO2 concentration in the upper 20 cm of the profile. δ13CCO2 became
slightly heavier in the upper 10 cm of the profile, with a maximum shift from −9.0 to −8.6‰ at 5 cm
depth. As at 14:30 h on Day 1, no significant surface CO2 fluxes were measured, thus the δ13CΔCO2 of
the surface flux could not be calculated from surface chamber data. Predicted surface flux rates
increased from 0.07 to 0.13 µmol m-2 s-1, with the predicted δ13CΔCO2 of the flux shifting from −17.4 to
−13.2‰. Soil temperature at 5 cm depth increased at close to its maximum rate, and soil
temperature at 15 cm depth increased from its daily minimum, reaching its maximum rate of change
by the end of the interval.
Day 2 (Figure 5.4 and Figure 5.5B)
Day 2 showed the same general patterns of behaviour in surface and subsurface CO2 dynamics, with
some differences in detail as described below. A step-by-step description of the surface and
subsurface CO2 dynamics is given in section 5.6.
Surface CO2 flux rates
On Day 1, predicted surface flux rates increased from low positive values at 14:30 h to a maximum at
22:30 h. Flux rates then decreased to negative values by 06:30 h before rising again to positive rates
by 10:30 h. The pattern of fluxes on Day 2 was similar, except for an anomalous negative predicted
flux at 22:30 h. On close examination, this negative flux is predicted as a consequence of a CO2
concentration at 0 cm depth which was greater than both the ambient soil and surface atmospheres.
The high CO2 concentration in the boundary layer must have been supported by CO2 production at
some shallow depth, which was not resolved by the relatively coarse (5 cm) sampling increments
employed. A concentration peak between 0 and 5 cm depth would have driven positive surface
fluxes.
116
Subsurface CO2 concentration and δ13CCO2 profiles
Between 14:30 and 18:30, while the soil was warming at 15 cm depth, the CO2 concentration profiles
remained similar, but δ13CCO2 values became more depleted at all depths. In the following 4-h period
to 22:30 h, CO2 concentration and δ13CCO2 profiles remained similar, other than an increase in
concentration (from 400−500 µL L-1) and associated depletion in δ13CCO2 (from −8.8 to −11.8‰) at 0
cm depth. This is similar to the increase in concentration and depletion in δ13CCO2 at 5 cm depth
between 18:30 and 22:30 h on Day 1.
The period from 22:30 to 02:30 h marked the greatest increase in CO2 concentration and decrease in
δ13CCO2 (up to 700 µL L-1 and −13.7‰, respectively, at 15 cm depth) while temperature was
decreasing throughout the profile. Between 02:30 and 06:30 h, subsurface CO2 concentrations
decreased from their maximum to below atmospheric CO2, and δ13CCO2 became more enriched, with
values close to that of atmospheric CO2. As for Day 1, the final 4-h period, from 06:30 to 10:30 h, was
marked by an increase in concentration throughout the profile to above atmospheric CO2
concentration in the upper 10 cm, but with no change in subsurface δ13CCO2 profile. As for Day 1,
significant surface CO2 flux rates were measured at the beginning and end of this 4-h period.
5.4 Discussion
Statistically significant surface CO2 flux rates at Site A were limited to 25% of the fluxes calculated
from surface chamber data. This limits the discussion of diel variation in surface CO2 flux rates to
those predicted from CO2 concentration gradients.
Predicted surface CO2 flux rates were highly variable, ranging from −0.18 to 0.33 µmol m-2 s-1 (Figure
5.4A). The duration and magnitude of positive surface CO2 flux rates exceeded that of negative flux
rates, and the δ13CΔCO2 of the fluxes (also predicted) was predominantly around −22‰, which is very
similar to the concentration-weighted δ13C of soil organic C in the upper 30 cm of the profile at Site A
(−22.8‰; Chapter 3, Figure 3.8). This pattern contrasts with that at Site B, where positive and
negative CO2 fluxes balanced, and the δ13CΔCO2 of the measured fluxes was highly variable, ranging
between −2.4 and −31.8‰ (Chapter 4, Figure 4.3). Furthermore, whereas at Site B, surface CO2 flux
rates varied in a sinusoidal manner with the same period, although not the same phase, as soil
temperature variation, the variability in predicted surface CO2 flux rates at Site A was not coupled to
soil temperature or the rate of change in soil temperature (Figure 5.6 and Figure 5.7).
117
Figure 5.6
Relationship between average temperature over the surface flux sampling period and
predicted surface CO2 flux rates at Site A, Taylor Valley. Regressions are not significant (p
> 0.1).
Figure 5.7
Relationship between the average rate of change in soil temperature over the surface
flux sampling period at 5 cm and 15 cm depth with predicted surface CO2 flux rates at
Site A, Taylor Valley. Regressions are not significant (p > 0.1).
At Site A, the highest surface CO2 flux rates and greatest increases in subsurface CO2 concentration
(and depletion in δ13CCO2) occurred at times when soil temperatures were cooling (22:30 and 02:30 h;
118
Figure 5.4 and Figure 5.5), thus eliminating exsolution of CO2 according to Henry’s Law as a possible
driver. Instead, the relatively high CO2 concentrations and associated depletion in δ13CCO2 values must
be due to biological processes. The highest predicted flux rate of 0.33 µmol m-2 s-1 is similar to rates
measured by Ball et al. (2009) at a site with slightly lower soil organic C content, but lower than rates
measured by Gregorich et al. (2006) in lakeshore soils with higher C and moisture levels. The finding
that the maximum rate of biological CO2 production occurred during periods when soil temperatures
were cooling runs counter to research which demonstrates an increase in biological activity with
increasing temperature (e.g. Lloyd and Taylor, 1994; Fang and Moncrieff, 2001). Nonetheless, this
evidence strongly suggests that diel CO2 dynamics at Site A were substantially influenced by
heterotrophic respiration, most likely fostered by the higher soil organic C and moisture levels at this
site. However, abiotic processes must still contribute, as shown by the negative CO2 flux rates that
occurred at 06:30 h, when soil temperatures were cooling and subsurface CO2 concentrations were
lower than that of atmospheric CO2.
The four-hourly sampling intervals, combined with the dynamic nature of subsurface CO2
concentration and δ13CCO2 profiles, precluded the identification of any steady-state phases.
Consequently, a mixing model approach could not be routinely applied to quantify the relative
contributions of biotic and abiotic processes to surface CO2 fluxes. In addition, the absence of steadystate phases means that the δ13CΔCO2 of predicted surface fluxes will have been affected by dynamic
fractionation effects (Amundson et al., 1998; Nickerson and Risk, 2009c), and thus cannot be directly
attributed to microbial utilisation of soil organic C, despite the δ13CΔCO2 values being close to that of
soil organic matter.
Nonetheless, there were two phases from which subsurface CO2 concentration and δ13CCO2 profiles
could be utilised, together with predicted surface flux δ13CΔCO2 values, to provide data suitable for
incorporation into a mixing model to constrain the contribution of biological production of CO2
within the soil. The first occurred between 10:30 h on Day 1, and the first sampling on Day 2, at 14:30
h. During this period, the soil produced CO2, and subsurface δ13CCO2 values became heavier (Figure
5.5). The surface CO2 flux was positive over this period, and the δ13CΔCO2 of the predicted surface CO2
flux shifted from −17.4 to −13.2‰ (Figure 5.4A and B). Assuming an intermediate isotopic
composition of −15.3‰ over this period, and combined with the fact that subsurface δ13CCO2 values
became heavier whilst CO2 concentration increased, the soil must have been producing CO2 with an
isotopic composition heavier than −15.3‰. For this condition to be satisfied, the mixing model
calculation (5.11) limits the biotic contribution to the flux to a maximum of 57%. This is based on the
assumption that 1) the δ13CCO2 of biologically-produced CO2 was −22.8‰, reflecting that of the source
organic material (Dörr and Münnich, 1980), and 2) the δ13CCO2 of abiotically-produced CO2 was
−5.2‰, as determined from steady-state periods of CO2 production at Site B.
119
The second phase that yielded data suitable for constraining biological contributions to surface CO2
fluxes occurred between 14:30 and 18:30 h on Day 2. During this period, subsurface CO2
concentration profiles remained steady whilst subsurface δ13CCO2 values became more depleted
(Figure 5.5B), and the δ13CΔCO2 of the predicted surface CO2 flux shifted from −13.2 to −19.6‰ (Figure
5.4B). The fact that there was no change in subsurface CO2 concentration whilst subsurface δ13CCO2
values became more depleted implies that the CO2 being produced in the soil must be isotopically
lighter than the surface flux over this period. Assuming the surface flux had an intermediate isotopic
composition of −16.4‰ throughout this interval, and the same assumptions as to the δ13CCO2 of
biologically- and abiotically-produced CO2 described above, the mixing model calculation (5.11)
requires that the proportion of biological CO2 production over this period is at least 64%. This period
represents steady-state with respect to CO2 production, and hence the predicted surface CO2 flux
rate equals the rate of soil CO2 production. Therefore, the biotic component of the flux is at least
64% of the predicted flux. Assuming an average CO2 flux of 0.1 µmol m-2 s-1 over this 4-h period
(Figure 5.4A), the biotic component to the total CO2 flux would be at least 0.06 µmol m-2 s-1.
The high spatial variability in subsurface CO2 concentrations and δ13CCO2 values during the cooler
periods of the day, when CO2 production was greatest, suggests that biological activity occurs in
discrete zones, perhaps controlled by highly localised organic C concentrations associated with
micro-scale landscape features (Moorhead et al., 2003). Alternatively, this variability may arise as a
result of pulses of CO2 from organisms responding to subtle and highly localised changes in soil
moisture conditions. An increase in the relative humidity within the soil as soil temperatures
decreased could be sufficient to provide additional moisture that may have activated biota (Büdel et
al., 2008), although any evidence of increased soil moisture content during these periods was not
detected by the Hydra Probes installed throughout the soil profile.
It is important to note that the periods of high production had a consistent pattern with respect to
soil temperature changes: peak CO2 production corresponded to the mid-point in soil temperature
decline at the depth at which CO2 was being produced. For example, the maximum CO2
concentration at 22:30 h on Day 1 was at 5 cm depth (Figure 5.5A); at this time, soil temperature at 5
cm depth was at the mid-point in its decline (Figure 5.4C). Similarly, the peak in soil CO2
concentration at 02:30 h on Day 1 was at 10 cm depth (Figure 5.5A), where soil temperature was
inferred to have been at the mid-point of its decline (Figure 5.4C). The CO2 peaks at 0 and 15 cm
depth at 22:30 and 02:30 h on Day 2, respectively, also corresponded to the mid-point in soil
temperature decline at those depths. This suggests that the soil biological community may be
responding to the combination of a decrease in temperature (to between ~3−5°C) and corresponding
increase in relative humidity. However, at temperatures below ~3−5°C, activity appears to be
inhibited.
120
5.4.1 Data reliability
The significant surface CO2 fluxes measured during low flux periods (06:30 and 10:30 h) were in good
agreement with those predicted using subsurface CO2 concentration gradients (Figure 5.4A). This
suggests that lateral diffusion and chamber-to-soil feedback effects during low flux periods were
negligible. However, the fact that not all of the surface CO2 fluxes calculated from these periods were
significant suggests that such data should be treated with caution. The limited number of statistically
significant measured surface fluxes also highlights the value of subsurface CO2 concentration profiles:
without them there is no means of validating surface CO2 flux rates.
At times when subsurface CO2 concentrations were above the ambient atmospheric CO2
concentration (400 µL L-1; at 14:30, 18:30, 22:30 and 02:30 h), surface CO2 flux rates were unable to
be derived from static chamber measurements because concentration changes were not systematic
over the 45 min sampling period (Figure 5.3). At Site A, where soils are characteristically dry and
porous, and biological activity is highly spatially and temporally variable, any non-linearities in
accumulation of CO2 in headspace chambers may be due to short-lived pulses of CO2 being produced
during the sampling period. The high spatial and temporal variability in biological activity is
highlighted in Figure 5.5, with relatively short-lived periods of biological CO2 production occurring
heterogeneously throughout the study area, often at times during which biological activity was
unexpected.
In addition, non-linearity in headspace chamber CO2 concentrations can be caused by chamber-tosoil feedbacks. Increases (or decreases) in CO2 concentration within the chamber distorts the
subsurface CO2 concentration gradient (Healy et al., 1996; Conen and Smith, 2000), inducing lateral
diffusion of CO2 (Creelman et al., 2013) as well as gas storage within the profile (Conen and Smith,
2000), thus causing underestimation of the surface CO2 flux rate. Despite the dry, porous nature of
the soil at Site A, lateral diffusion effects were minimised by installing the lowermost edges of the
chambers to 6 cm below the soil surface, and maximising their height above the soil surface as much
as was practicable given the expected flux rate (Healy et al., 1996; Conen and Smith, 2000). However,
as the expected surface CO2 flux rate, based on data obtained in the 2008/09 field season (Chapter 3,
Table 3.1), was low, the 45 min chamber deployment would likely have exacerbated lateral diffusion
effects during periods of comparatively high (> 0.05 µmol m-2 s-1) CO2 fluxes. A shorter deployment
period during times of high flux may have reduced lateral diffusion effects, and therefore increased
the likelihood of measuring statistically significant surface CO2 flux rates.
Chamber-to-soil feedback effects commonly manifest as quadratic or exponential growth curves of
CO2 concentration (Creelman et al., 2013). However, no such monotonic increases in CO2
concentration were observed. Despite the variability in initial CO2 concentrations within the
121
chambers, the non-linearities were not systematic (Figure 5.3), which suggests that the changes in
CO2 concentration may have been a result of highly temporally (and spatially) variable CO2
production, rather than a chamber-induced feedback effect. Furthermore, the inflections in the timeseries data (Figure 5.3) would require CO2 to be diffusing out of the chambers at times. Diffusion into
the soil is unlikely given the pervasive positive concentration gradients that existed during all
sampling periods other than that at 06:30 h. Wind pumping effects are possible, but they were not
observed during simultaneous sampling periods at Site B, where wind speeds would have been equal
to or greater than those measured at Site A. Such effects cannot be evaluated for Site A, as surface
CO2 flux rates were predicted from subsurface CO2 concentration gradients. The observed nonlinearities could also have arisen from analytical error, although this is considered unlikely. Samples
from Sites A and B were analysed in the same runs, on multiple days, and there was no evidence of
analytical error. The cause of the inflections remains unknown.
5.4.2 Implications
Despite the lack of significant surface CO2 fluxes throughout most of the 48-h sampling period, diel
variation in subsurface CO2 concentrations and δ13CCO2 at Site A clearly demonstrates that production
of CO2 does not occur at a constant rate throughout the 24-h cycle, nor does the maximum rate of
CO2 production correspond to maximum soil temperatures. Instead, the maximum rate of CO2
production at Site A (inferred from subsurface CO2 concentration and δ13CCO2 profiles) occurred
during periods when soil temperatures were low and decreasing. This implies that soil biological
activity at this site is influenced by interactions between temperature and moisture availability.
Although not a surprising result in itself, the inference that subtle changes in relative humidity,
driven by decreases in soil temperature, suggests that other abiotic factors known to influence
biological activity (e.g. soil geochemistry; Barrett et al., 2004; Poage et al., 2008) are of lesser
importance at this site. The assumption that some microorganisms may be able to directly absorb
water vapour (Cary et al., 2010; Stomeo et al., 2012) supports the idea that the biological community
at this site may comprise organisms that are able to respond to subtle changes in relative humidity.
The finding that CO2 production is non-steady, with peak production during cooler parts of the day,
reinforces the significance of the results presented in this study and highlights problems that were
not identified during previous studies of Dry Valley soil CO2 fluxes. It can no longer be assumed that:
1) surface CO2 fluxes are solely biological in origin, 2) the maximum rate of CO2 production occurs
during the warmest parts of the day, and 3) surface CO2 flux rates are representative of soil CO2
production (i.e. the system is in steady-state). An understanding of both surface and subsurface CO2
dynamics is vital for identifying periods of steady-state, during which time surface CO2 fluxes and
122
their isotopic composition are representative of CO2 production within the soil (Amundson et al.,
1998; Nickerson and Risk, 2009c). It is only during such periods that a valid assessment of the relative
contributions of biotic and abiotic processes to surface CO2 fluxes can be made.
Previous studies of soil CO2 flux rates in the McMurdo Dry Valleys have not differentiated between
biotic and abiotic components of the flux. Furthermore, it has been assumed that maximum surface
CO2 flux rates, measured at the warmest times of the day, represent maximum rates of biological
respiration (Burkins et al., 2001; Ball et al., 2009). Having demonstrated that biological activity
appears to be greatest during periods of low and decreasing soil temperatures at Site A, and that soil
CO2 production is non-steady, any inferences about soil respiration rates that are based solely on
surface flux data should be re-evaluated. Further discussion of the implications of these results on
interpretation of surface soil CO2 fluxes measured in Dry Valley soils will be presented in the
following chapter.
The limited number of significant surface CO2 flux rates measured at Site A highlights the importance
of tailoring sampling protocols to suit expected flux rates at each site. The higher organic C and soil
moisture content at Site A is likely to support a more active biological community than that at Site B,
away from any localised sources of C or moisture. In retrospect, the sampling period at Site A should
have been shortened in anticipation of greater biological activity, although data from the 2008/09
field season (Chapter 3) were wrongly assumed to have captured the maximum variability in surface
CO2 flux rates. The fact that they did not means that assumptions made in Chapter 3 are
inappropriate with respect to the inferred maximum biological contribution to surface CO2 fluxes.
The surprising result that maximum CO2 production occurred during periods when soil temperatures
were low and decreasing meant that the 45-min sampling periods were probably too long to produce
reliable surface chamber data. However, despite the fact that data from the 45-min sampling periods
at Site A largely failed to show a consistent increase (or decrease) in CO2 concentration with time, the
non-systematic variability in CO2 concentrations suggested that temporal variability in CO2
production, rather than lateral diffusion of CO2 in response to distorted subsurface concentration
gradients, limited the calculation of surface CO2 flux rates. Predictions of surface CO2 flux rates based
on consistent subsurface CO2 concentration profiles appeared reliable during low flux periods, but
high spatial variability in subsurface CO2 concentration profiles during cooler parts of the day limited
their reliability for predicting surface CO2 flux rates at times of high flux (> 0.10 µmol m-2 s-1).
Nonetheless, the spatial variability in subsurface CO2 concentrations suggests that any biological
component of soil CO2 fluxes is unlikely to be homogenous, even throughout a relatively small
sampling area. Observed variability in surface CO2 flux rates may be in response to the distribution of
organic C or to localised sources of moisture, or a combination of both factors. As such, any study
123
attempting to differentiate between biotic and abiotic components of surface CO2 fluxes cannot
discount the possibility that within a sampling area, some fluxes may have a negligible biological
component. Studies combining subsurface CO2 concentration and δ13CCO2 profiles with surface flux
measurements are vital in order to identify periods of steady-state, during which surface CO2 fluxes
and their isotopic composition are representative of CO2 production within the soil. In addition,
subsurface CO2 concentration and δ13CCO2 profiles enable identification of the depth at which
biological CO2 production is occurring.
5.5 Summary and conclusions
The diel changes in subsurface CO2 concentration and δ13CCO2 profiles described in this chapter are
likely to be representative of subsurface CO2 dynamics in Dry Valley soils with relatively high pH, low
EC, and above-average organic C and moisture content. In these soils, subsurface (and surface) CO2
dynamics are driven by a combination of biotic and abiotic processes, with the relative proportions of
each determined by trends in soil temperature. The above-average organic C and moisture content in
soils at Site A sustained biotic contributions that were undetectable at Site B, where pH and EC were
similar, but organic C and moisture content were considerably lower.
Despite the inability to calculate surface CO2 fluxes from chamber measurements and the lack of
periods of steady state at Site A, and therefore a means by which to quantify the relative proportions
of biotic and abiotic components of surface CO2 fluxes, the subsurface CO2 dynamics provided a basis
for semi-quantitative interpretation of the timing and relative contribution of biological CO2
production. Subsurface CO2 concentration and δ13CCO2 profiles in this relatively moist and high C
environment were highly spatially and temporally variable, and, most significantly, it appears that
biological production of CO2 was greatest between 5 and 20 cm depth, at times when soil
temperatures were low and decreasing. At these times (22:30−02:30 h), soil CO2 production and
surface CO2 fluxes were most likely entirely of biological origin. Previous interpretations of surface
CO2 flux rates in the Dry Valleys, which assumed that maximum surface flux rates occurred during the
warmest part of the day and represented the maximum rate of biological activity, should be reevaluated for two main reasons. Firstly, surface CO2 flux rates are highly variable, comprising both
biotic and abiotic components, and the maximum rate of CO2 production, at least at this site,
occurred during cooler parts of the day. Secondly, as CO2 production is non-steady, surface CO2 flux
rates are not representative of the rate of CO2 production within the soil.
The generality of these conclusions and their relationship to data from Site B, along with further
discussion of diel variability in surface CO2 fluxes and implications for soil respiration measurements
in the McMurdo Dry Valleys is considered in the following chapter.
124
5.6 Supplementary information
5.6.1 Detailed description of surface and subsurface CO2 dynamics on Day 2
The following descriptive summaries of surface and subsurface CO2 dynamics on Day 2 relate to data
presented in Figure 5.4 and Figure 5.5B.
14:30 – 18:30 h
Over this interval, the subsoil continued to accumulate a small amount of CO2. Increases in CO2
concentration occurred beneath 10 cm depth, and by 18:30 h, CO2 concentrations were greater than
that of atmospheric CO2 throughout the profile. Despite the relatively small increases in CO2
concentration, δ13CCO2 became significantly more depleted throughout the profile, with a maximum
shift from −8.4 to −9.5‰ at 10 cm depth. No significant surface CO2 fluxes were measured at 14:30
or 18:30 h, thus the δ13CΔCO2 of the flux could not be calculated from surface chamber data. Predicted
surface flux rates decreased from 0.13 to 0.07 µmol m-2 s-1, and the δ13CΔCO2 of the flux became more
depleted, shifting from −13.2 to −19.6‰.
Soil temperatures at 5 and 15 cm depth reached their daily maxima during this interval. However,
soil temperature at 5 cm depth reached its maximum early in the interval and then decreased, at
times at a rate close to its maximum negative rate of change, whereas soil temperature at 15 cm
depth increased throughout the interval, initially at its maximum rate, reaching its daily maximum at
the end of the interval. The increase in subsurface CO2 concentrations and the depletion in soil
δ13CCO2 and predicted surface flux δ13CΔCO2 values over this interval is similar to that which occurred
during the 14:30 to 18:30 h interval on Day 1. Soil temperatures at 5 and 15 cm depth during the
14:30 to 18:30 h interval on Day 1 were also similar to those measured during this interval.
18:30 – 22:30 h
Between 18:30 and 22:30 h, subsurface CO2 concentrations increased in the upper 5 cm of the
profile, and decreased slightly between 5 and 20 cm depth, although concentrations remained
greater than that of atmospheric CO2 concentration throughout the profile. However, there was no
net change in subsurface CO2 concentration. Subsurface δ13CCO2 values became significantly more
depleted in the upper 5 cm of the profile, and slightly more depleted throughout the remainder of
the profile. Surface CO2 fluxes remained non-significant, perpetuating the inability to determine the
δ13CΔCO2 of the net CO2 flux from surface chamber data. Predicted surface CO2 fluxes decreased from
0.07 to −0.18 µmol m-2 s-1. However, the negative predicted flux at 22:30 h is spurious, as the high
CO2 concentration in the boundary layer, as measured by the 0 cm sampling tube (Figure 4.4B),
means that the concentration gradient into the soil is negative, yet there is a positive CO2
concentration gradient between the boundary layer and the turbulent atmosphere above. This
125
situation probably arises from CO2 production at a shallow soil depth, which is unable to be resolved
by the sampling strategy employed. The change in CO2 concentration and δ13CCO2 gradients acted to
switch surface fluxes from a relatively depleted predicted efflux (−19.6‰) at 18:30 h to a more
depleted predicted influx (−24.5‰) at 22:30 h, although the high CO2 concentration at 0 cm depth at
22:30 h means that this value is unreliable. Soil temperature at 5 cm depth decreased, at times at
close to its maximum negative rate, whereas soil temperature at 15 cm depth gradually decreased
from its daily maximum. The soil temperature trends observed during this period were similar to
those that occurred between 18:30 and 22:30 h on Day 1. Subsurface CO2 concentration and δ13CCO2
profiles were also similar, although predicted surface CO2 flux rates were substantially greater on Day
1 than Day 2.
22:30 – 02:30 h
During this period, the soil gained CO2 at its maximum rate of 0.05 µmol m-2 s-1. Subsurface CO2
concentrations increased significantly throughout the profile, with a maximum increase from 420 to
700 µL L-1 at 15 cm depth. Significant depletions in subsurface δ13CCO2 values accompanied the
increases in CO2 concentration; the maximum depletion in δ13CCO2 was at 15 cm depth, shifting from
−9.2 to −13.7‰. The profile average CO2 concentration (560 µL L-1) and δ13CCO2 (−12.6‰) at 02:30 h
were the highest and most depleted, respectively, recorded throughout the 48-h sampling period.
However, surface CO2 fluxes were again not significant, thus the δ13CΔCO2 of the net CO2 flux could not
be calculated from surface chamber data. The predicted surface flux was 0.24 µmol m-2 s-1 at 02:30 h,
and changes in the CO2 concentration and δ13CCO2 gradients drove a relatively depleted predicted
efflux (−20.9‰) at this time. Soil temperatures at 5 and 15 cm depth continued to decrease; the rate
of change in soil temperature at 15 cm depth was close to its maximum negative rate. The significant
increase in subsurface CO2 concentrations and depletion in δ13CCO2 was considerably greater than
that which occurred during the 22:30 to 02:30 h interval on Day 1, despite soil temperatures at 5 and
15 cm depth being similar on both days.
02:30 – 06:30 h
Over this interval, the subsoil lost CO2 at its maximum rate of −0.07 µmol m-2 s-1 as CO2
concentrations decreased from their maximum to their minimum. By 06:30 h, subsurface CO2
concentrations were less than that of atmospheric CO2 concentration throughout the profile, with
the greatest decrease from 700 to 350 µL L-1 at 15 cm depth. Subsurface δ13CCO2 values became
significantly more enriched throughout the profile, with the maximum enrichment in δ13CCO2 (up to
5.1‰) occurring between 15 and 20 cm depth. At 06:30 h, one significant positive flux and one
significant negative flux were measured, suggesting that surface CO2 fluxes were spatially variable at
this time (note that Figure 5.4 shows the average significant flux), although the surface flux predicted
from the CO2 concentration gradient at 06:30 h was −0.04 µmol m-2 s-1. The δ13CΔCO2 of the net CO2
126
gained in the surface chamber with the significant positive flux (0.07 µmol m-2 s-1) was −6.6‰,
whereas the δ13CΔCO2 of the net CO2 lost in the surface chamber with the significant negative flux
(−0.03 µmol m-2 s-1) was 5.2‰. The significant changes in CO2 concentration and δ13CCO2 throughout
the profile resulted in a relatively heavy predicted influx of CO2 (−8.0‰). Soil temperatures at both 5
and 15 cm depth decreased during this interval, with the latter decreasing at its maximum negative
rate. The dramatic changes in subsurface CO2 concentration and δ13CCO2 profiles were similar to those
observed during the 02:30 to 06:30 h interval on Day 1, although the changes during this interval
were considerably greater. Changes in soil temperature during the 02:30 to 06:30 h intervals on both
Days 1 and 2 were almost identical at 15 cm depth, but soil temperature at 5 cm depth decreased
from 2.2 to −0.6°C on Day 2, as opposed to remaining relatively constant around 1.5°C on Day 1.
06:30 – 10:30 h
From 06:30 h to the final sampling at 10:30 h, the soil returned to net accumulation of CO2.
Subsurface CO2 concentration increased throughout the profile, and by 10:30 h, CO2 concentrations
were greater than that of atmospheric CO2 concentration in the upper 10 cm of the profile. Despite
the increase in subsurface CO2 concentrations, δ13CCO2 values did not change significantly. Significant
surface CO2 fluxes were recorded in all four surface chambers, averaging 0.04 µmol m-2 s-1. This was
similar to the predicted surface CO2 flux rate at 10:30 h (0.06 µmol m-2 s-1). The average δ13CΔCO2 of
the net CO2 flux in the surface chambers was −13.0‰; this was identical to that predicted based on
CO2 concentration and δ13CCO2 gradients at 10:30 h. Soil temperature at 5 cm depth decreased to its
daily minimum early in the interval, before increasing at its maximum rate. Soil temperature at 15 cm
depth decreased, reaching its daily minimum in the second half of the interval. The increase in
subsurface CO2 concentrations, significant positive surface CO2 flux rates and relatively enriched
δ13CΔCO2 of the surface flux, along with changes in soil temperature observed over this period were
similar to those that occurred during the 06:30 to 10:30 h interval on Day 1.
127
128
Chapter 6
Synthesis and Summary
6.1 Introduction
The Dry Valleys of Antarctica are home to life at the limit, where the terrestrial environment is
characterised by short summers, long dark winters, and extremes in temperature and wind.
Precipitation is extremely limited, and moisture availability is controlled by temperature, which
dictates availability of meltwater from glaciers and snow patches, as well as melting of ice-cemented
permafrost, during short periods in the austral summer when temperatures are above freezing.
Despite the generally high salinity and low levels of C, N and moisture in Dry Valley soils, biological
communities are widely distributed throughout the landscape. However, biological activity tends to
be concentrated in “hotspots” where nutrients are adequate, salinities are relatively low, and
temperatures during the austral summer are sufficiently warm to sustain liquid water.
There remains considerable interest in understanding the structure and function of the communities
that survive within the harsh Dry Valleys environment. As well as allowing fundamental questions of
life at the limit to be addressed, which is of great interest for astrobiology, the resilient communities
inhabiting Dry Valley terrestrial ecosystems, where there are no vascular plants, allow for trophic
interactions to be examined in a low-complexity system (Hogg et al., 2006). Whilst early research
concluded that biological diversity was low, modern sequencing techniques have shown that the
distribution and diversity of organisms inhabiting the Dry Valleys is considerable, with abiotic factors
exerting the dominant control on community structure and function (Lee et al., 2012; Stomeo et al.,
2012; Van Horn et al., 2013).
One major focus of Dry Valley terrestrial ecosystem research has been identifying the sources of C
that sustain these communities. Two models, which are not necessarily mutually exclusive, have
been proposed. The first to be suggested was a “legacy” model, whereby ecosystems are primarily
sustained by lacustrine-derived organic C inherited from primary production within large paleolakes
(Burkins et al., 2000). An alternate explanation, the “spatial subsidy” hypothesis, proposes that
contemporary, wind dispersed lacustrine-derived organic C is sufficient to sustain modern ecosystem
activity (Elberling et al., 2006). However, this model may be of limited applicability to the broader Dry
Valleys region, as it requires a relatively high source (primary production zone – i.e. lakes) to sink
(area which is devoid of primary productivity zones) ratio, which is uncommon.
129
Support for both hypotheses has come from studies investigating soil respiration rates and C
turnover times (Burkins et al., 2001; Barrett et al., 2005, 2006a; Elberling et al., 2006), as well as
studies using C and N isotopes to define the provenance of the source organic materials (Burkins et
al., 2000; Hopkins et al., 2009). Previous studies of in situ soil respiration rates have implicitly
assumed that surface CO2 fluxes are representative of CO2 production within soils. This requires
steady-state conditions, which, until now, have not been tested, despite several studies showing that
surface CO2 flux rates are highly variable (Parsons et al., 2004; Ball et al., 2009). The diel variability in
surface CO2 flux rates, documented by Parsons et al. (2004) and Ball et al. (2009), extended to
negative CO2 fluxes during parts of the day. As no biological explanation can be invoked to explain
the CO2 influxes, Parsons et al. (2004) and Ball et al. (2009) suggested that there was an abiotic
component to the fluxes that could be explained by dissolution and exsolution of CO2 in accordance
with Henry’s Law.
This current study aimed to test the hypothesis of abiotic CO2 dissolution and exsolution using stable
C isotope signatures to determine what contribution abiotic processes can make to surface CO2
fluxes in Dry Valley soils.
6.2 Key findings
This study focused on two sites with contrasting soil organic C contents in eastern Taylor Valley. Soils
at both sites were coarse textured, formed in glacial drift, and classified as Typic Haploturbels. Site A
was located near a small lake, and can be considered a “biological hotspot”, with the lake providing a
contemporary organic C source via algal growth during summer months. Subsequent wind dispersal
of algae accumulated on the lake shore is inferred to provide a local source of organic C to nearby
soils. Interestingly, as well as the δ13C values of soil organic C at Site A being typical of a lacustrinederived source, C isotopic signatures characteristic of endolithically-derived organic matter (Burkins
et al., 2000; Hopkins et al., 2009) were evident between 5 and 15 cm depth (Chapter 3, Figure 3.8). In
contrast, Site B was distal from any contemporary lacustrine inputs, thus limiting the source of
organic C to local inputs via spalling of rock containing endolithic communities (Mergelov et al.,
2012), and “legacy” C from large paleolakes that existed during the Last Glacial Maximum.
Alternatively, it could be argued that Site B receives inputs of organic C via spatial subsidies, although
this is unlikely considering the small source-to-sink ratio in Taylor Valley. The only substantial
contemporary sources of organic C in Taylor Valley are from moats around Lakes Bonney, Hoare, and
Fryxell, as well as smaller lakes such as at Site A, and other ephemeral ponds and streams
(Moorhead, 2007); these zones of productivity represent only a small proportion of the total
landscape. Furthermore, variation in the isotopic signature of soil organic C with elevation in Taylor
130
Valley suggests limited transport of organic C in the landscape (Burkins et al., 2000). The δ13C profile
of soil organic C at Site B implies that endolithically-derived C overlies low levels of lacustrine-derived
organic matter (Chapter 3, Figure 3.8).
Chapter 3 aimed to verify the observations of Parsons et al. (2004) and Ball et al. (2009) by examining
subsurface CO2 concentration and δ13CCO2 profiles measured during the 2008/09 austral summer.
Concentration profiles measured at the warmest and coolest times of the day were consistent with
positive and negative surface CO2 fluxes, respectively. A physical model, capturing the temperaturedriven dissolution and exsolution of CO2, which utilised empirical data (soil moisture content, pH, and
temperature changes measured in the field), provided an adequate explanation of the observed
subsoil CO2 dynamics. The model assumed equilibrium partitioning of dissolved inorganic C species
and a closed system with respect to CO2 fluxes, and this study, like previous studies of soil
respiration, was predicated on the assumption that the maximum rate of CO2 production occurred
during the warmest part of the day. A two-component mixing model, based on the assumption that
the δ13CCO2 of biologically-produced CO2 was the same as the δ13C of the source organic C, and a range
of assumptions as to the δ13CCO2 of abiotically-produced CO2, constrained the biological contribution
to subsurface CO2 production (between 5 and 15 cm depth) at Site A to < 25%.
The results from the first study (Chapter 3), based on a twice-daily sampling regime, did not resolve
diel changes or rates of change in surface CO2 fluxes and subsurface CO2 concentration and δ13CCO2
profiles. Furthermore, it was not known if steady-state phases occurred, from which the rate of soil
CO2 production could be estimated from surface CO2 flux rates. These questions formed the basis for
the sampling strategy implemented during the second (2009/10 austral summer) field season.
During the 2009/10 austral summer, a 48-h time-series of surface CO2 fluxes and subsurface CO2
concentration and δ13CCO2 profiles was measured at 4-hourly intervals at Site B (Chapter 4). The
surface CO2 flux rates and subsurface CO2 concentration and δ13CCO2 profiles measured at the
warmest and coolest times of the day were consistent with measurements made in the previous
season. However, the increased sampling frequency revealed the dynamic behaviour of subsurface
CO2 concentration and δ13CCO2 as soil temperatures varied throughout the diel cycle. The subsurface
CO2 concentration gradients were consistent with the direction of the surface CO2 flux, with positive
and negative concentration gradients driving positive and negative surface CO2 fluxes, respectively.
Changes in surface CO2 flux rates and subsurface CO2 storage flux rates were statistically significantly
related to the rate of change in subsurface soil temperature. These results support Henry’s Lawcontrolled CO2 exsolution and dissolution driving production and consumption of CO2, respectively. In
addition, the 13C isotopic composition of the net surface CO2 fluxes (δ13CΔCO2) could be reliably
predicted from subsurface CO2 concentration and δ13CCO2 gradients, confirming that the approach
131
used for calculating the δ13CΔCO2 of measured surface fluxes was appropriate. Keeling plots could not
be used because of the small changes in CO2 concentration in the chambers over the 45 min
sampling interval. The relationship between surface CO2 flux rate and near-surface CO2 concentration
gradients also allowed an effective diffusivity to be calculated.
The variability in subsurface CO2 concentration and δ13CCO2 profiles demonstrated that CO2
production was characteristically non-steady. Over the 48-h sampling period, CO2 concentration and
δ13CCO2 profiles varied symmetrically about ambient atmospheric CO2 concentration and δ13CCO2
(Chapter 4, Figure 4.4). The non-steady conditions induced dynamic fractionation, which resulted in
highly depleted surface CO2 fluxes (as low as ~ −30‰) that could not be attributed to heterotrophic
respiration from organisms utilising a highly depleted organic C source. Where short periods (4 h) of
steady-state efflux were identified, they showed CO2 production (exsolution) of relatively enriched
CO2 (−5.2‰); periods of steady-state influx showed consumption (dissolution) of lighter CO2
(−11.4‰). This pattern is consistent with the kinetic fractionation expected as a result of preferential
exsolution of 13CO2 and preferential dissolution of 12CO2 in the CO2 (g) – CO2 (dissolved) system. Despite
the relatively low surface CO2 flux rates measured at Site B, the magnitude of surface fluxes was large
relative to subsurface storage CO2 fluxes. This finding confirmed that the closed system assumption
used in Chapter 3 was an over-simplification, which most significantly affected comparisons between
measured and simulated ratios of subsurface CO2 concentration in the near-surface zone (0−5 cm), as
speculated in Chapter 3. Together, the results of Chapter 4 confirm that at Site B, with low levels of
soil organic C and no significant contemporary source of organic C, there was no detectable biological
contribution to CO2 production.
Chapter 5 presents the same suite of data as at Site B, measured concurrently and at the same
frequency so as to provide insights into CO2 dynamics at a contrasting biological “hotspot” (Site A).
The aim of this chapter was to test the generality of the findings at Site B at a site with a
contemporary organic C supply, which can be considered a “biological hotspot” (Site A). As at Site B,
the CO2 concentration and δ13CCO2 profiles measured at the warmest and coolest times of the day
were similar to those measured during the previous season. However, the increased temporal
resolution of subsurface CO2 concentration and δ13CCO2 profiles revealed contrasting behaviour to
that at Site B. At 06:30, 10:30 and 14:30 h, the CO2 concentration and δ13CCO2 profiles resembled
those at Site B. However, at 18:30, 22:30 and 02:30 h, CO2 concentrations were well above those
measured at Site B, reaching a maximum at 02:30 h; the most depleted δ13CCO2 values were also
measured at this time. The high CO2 concentrations and relatively depleted δ13CCO2 values imply
biological production of CO2, despite occurring at a time when the soil was cooling, although soil
temperatures remained above freezing. Consequently, it is suggested that the high CO2
concentration and associated depletion in δ13CCO2 was a result of cooling-induced condensation,
132
providing soil biota with liquid water, or at least high humidity, sufficient to stimulate their activity.
This inference suggests that the assumption that the maximum rate of CO2 production occurs during
the warmest part of the day, as used in Chapter 3 and in previous studies of Dry Valley soil
respiration, is not appropriate for all sites.
Surface CO2 flux rates at Site A could not be reliably calculated, as CO2 concentrations varied nonsystematically throughout the 45-min chamber sampling periods. Consequently, subsurface CO2
concentration gradients, together with a value for effective diffusivity of CO2 derived from Site B,
were used to estimate surface CO2 flux rates. The δ13CΔCO2 of surface CO2 fluxes was predicted from
CO2 concentration and δ13CCO2 gradients, but despite often being similar to the δ13C of the soil organic
C, the non-steady CO2 dynamics (and resulting dynamic fractionation) limit the interpretations that
can be made as to the biological contribution to the flux to semi-quantitative estimates. Nonetheless,
the changes in subsurface CO2 concentration and δ13CCO2 profiles are consistent with the inference
that heterotrophic respiration is an important, and at times, dominant component of subsurface CO2
production and surface CO2 fluxes at Site A. The CO2 dynamics at this site are likely to be
representative of “biological hotspots” in the Dry Valleys, where organic C and liquid water are less
limiting than in other parts of the landscape.
6.3 Implications of research findings
The diel variability in surface CO2 flux rates and subsurface CO2 concentration and δ13CCO2 profiles has
important implications for the interpretation of Dry Valley soil CO2 flux measurements. The results
presented in this study are particularly relevant to studies investigating ecosystem activity and
response to environmental change, and C turnover times in the Dry Valleys, as well as CO2 flux
studies in other deserts (e.g. Ma et al., 2013; Brummell et al., 2014).
Previous studies of Dry Valley soil CO2 fluxes have not considered surface and subsurface CO2
dynamics in such detail, and have based their interpretations of surface CO2 flux rates on the
following two assumptions: 1) that maximum surface CO2 flux rates occur at the warmest times of
the day and represent the maximum rate of soil CO2 production, and 2) that surface CO2 fluxes are
representative of CO2 production within the soil (i.e. the system is in steady-state). This study has
demonstrated that these two assumptions do not hold for all sites, hence new approaches to
characterisation of Dry Valley soil CO2 fluxes are necessary.
133
6.3.1 Comparisons of Dry Valley ecosystem activity
The misinterpretation of surface CO2 flux rates as representing the rate of CO2 production within the
soil is problematic for studies comparing the timing and magnitude of ecosystem activity between
sites. Firstly, the results presented in this study have demonstrated that Dry Valley CO2 dynamics are
characteristically non-steady. Therefore, the surface CO2 flux is not representative of soil CO2
production at the time of measurement. Characteristic response times (t) for diffusive systems are
related to length scales (L) and diffusivity (D) by t = L2/D (Anderson and Anderson, 2010). For the
effective diffusivity of the coarse-textured soils studied (0.03 cm2 s-1), perturbations in CO2
production at 15 cm depth would take ~2 h to affect surface fluxes. Therefore, periods of steadystate production that persist for up to 2 h are probably necessary to yield steady surface CO2 flux
rates. This appears to occur in soils where biological CO2 production is low (Site B), but CO2 dynamics
in soils where biological CO2 production is relatively high (Site A) are characterised by high spatial and
temporal variability in subsurface CO2 production, hence steady-state is rarely observed. Secondly,
site-specific differences in the timing of maximum CO2 production are likely to be regulated by local
soil temperature and moisture availability, as well as the functionality of the biological community
present. This means that local site conditions must be carefully monitored to ensure comparability
between datasets. Finally, this study shows that surface CO2 flux rates alone do not provide an
indication as to the relative proportions of biotic and abiotic contributions to the flux. Consequently,
comparisons between ecosystem activity based on surface CO2 flux rates may be misleading.
6.3.2 Carbon turnover times
Estimates of C turnover times are contingent on reliable measures of heterotrophic soil respiration.
The caveats discussed above suggest that existing estimates of C turnover times may be erroneous.
At sites where there is a substantial abiotic component to the flux, turnover times calculated on the
basis that soil CO2 fluxes represent biological CO2 production will be underestimates. Any abiotic
component to the flux means that C utilisation rates will be less than previously assumed, hence C
pools will have a longer turnover time. Alternatively, at sites with high biological activity, turnover
times could be underestimates if the surface CO2 flux rates on which they are based did not capture
flux rates at the time of maximum CO2 production. This has implications for the legacy and spatial
subsidy models of C supply. Current estimates of C turnover time, based on in situ CO2 flux rates,
range from 23 to 130 years (Burkins et al., 2001; Elberling et al., 2006), which is incompatible with
the legacy model. However, it has been suggested that there may be at least 2 pools of organic C
involved in Dry Valley C cycling: a small, labile organic C pool which is sustained via in situ autotrophic
productivity, and a larger, more recalcitrant organic C pool derived from ancient glacial and lake
sediments (Burkins et al., 2001; Barrett et al., 2005). Regardless of the organic C pools involved, if soil
134
CO2 flux rates used in calculations of C turnover times are overestimates, then current ecosystem
activity may in fact be able to be sustained by legacy C.
6.3.3 Field and laboratory manipulations
The results presented in this study can also be used to inform laboratory studies that utilise Dry
Valley soils. As with temperate soils, laboratory studies of Dry Valley soil microbial activity also show
a greater respiratory response to increased soil temperatures (Elberling et al., 2006; Hopkins et al.,
2006). However, most of these studies also increased moisture content and provided additions of C
and N (Hopkins et al., 2006; Hopkins et al., 2008; Sparrow et al., 2011), thereby providing favourable
conditions for biological activity. No studies have attempted to simulate the diel temperature
variability that characterises field conditions, instead maintaining a relatively high but stable
temperature equivalent to the maximum summertime soil temperature (Hopkins et al., 2006;
Hopkins et al., 2008; Sparrow et al., 2011). Without temperature variability, there is no mechanism
for abiotic CO2 production, and thus on the basis of increased respiration rates under consistently
warm temperatures (and following substrate addition), these studies have concluded that soil CO2
production is of biological origin. However, without a genuine simulation of field conditions, results
from subsidy experiments do not provide sufficient evidence to preclude abiotic CO2 production
under field conditions. Similarly, laboratory decomposition studies that show δ13C values of emitted
CO2 are similar to the δ13C values of source organic materials (Hopkins et al., 2009) simply confirm
the inference of a biological response to increased temperature and substrate addition. Again, this
would be expected in a laboratory environment, but it cannot be used to infer that biological
processes dominate CO2 production in the field. As steady-state (with respect to temperature)
laboratory experiments cannot simulate the range of processes occurring in the field, such studies
will generally eliminate abiotic processes. Consequently, conclusions from studies conducted at a
constant temperature must be re-evaluated before being assumed to fully describe CO2 dynamics in
the field.
Field manipulations such as those incorporating substrate additions, and artificially increasing soil
temperature and moisture content, will also benefit from insight gained through this study. In
particular, the timing and frequency of soil CO2 flux measurements is critical, and may significantly
influence the interpretations made. Further insight could be readily gained from studies such as
Barrett et al.’s (2008a) tracer study, which used a 13C-labelled sugar to track heterotrophic CO2
production from nematodes. If the measured surface CO2 fluxes (from which the nematode
contribution to surface fluxes was calculated) included an abiotic component, then the contribution
to biological respiration by nematodes could be significantly more than the ~7% calculated.
135
6.4 Future research
This study is the first study that has attempted to characterise subsurface soil CO2 dynamics in the
Dry Valleys. The data presented in this thesis therefore represent a first attempt at quantifying the
biotic and abiotic contributions to surface CO2 fluxes at two sites that represent two of the major soil
ecosystems in the Dry Valleys region. As a result of its exploratory nature, this study has led to
various questions that must yet be addressed in order to develop a better understanding of Dry
Valley soil C cycling, which will subsequently contribute to knowledge used in predicting the
implications of changing environmental conditions on soil C dynamics throughout the Dry Valleys.
6.4.1 Laboratory studies
Firstly, and arguably most importantly, the δ13CCO2 of abiotically-produced CO2 must be empirically
derived. This will remove the need for imprecisely-constrained or assumed values of δ13CCO2-abiotic in
mixing model calculations. This value can be obtained by sampling soil CO2 from cores within “Dry
Valley simulators”, which, as the name suggests, can simulate the diel temperature variability
experienced in Dry Valley soils. As fractionation during exsolution of CO2 is kinetically-driven,
experiments that replicate rates of soil temperature change at different soil depths or at different
times of the year would be informative.
Metagenomic studies of the biological communities at different depths at Sites A and B will also
enable questions about community structure and function to be addressed; subsequent studies in a
controlled environment in which abiotic factors are varied will facilitate a better understanding of
the controls on the biological community function at each site. As most studies of Dry Valley soil
biological communities focus on samples from the top 10 cm of the profile, characterisation of the
communities present between 10 and 20 cm depth may reveal unexpected differences in community
structure. This is of particular relevance at Site A, as biological CO2 production appears to be greatest
between 10 and 20 cm depth at cooler times of the day.
6.4.2 Numerical modelling studies
Modelling studies have not been incorporated into this study as several key parameters (the δ13CCO2
of abiotically-produced CO2, and the rates of biotic and abiotic CO2 production) needed to inform
realistic simulations have not yet been quantified. However, once these parameters are obtained,
much insight into soil CO2 dynamics under differing soil physical and chemical conditions could be
gained from modelling studies.
Physical models (based on field data), such as those developed by Nickerson and Risk (2009c) to test
lag responses and dynamic fractionation effects resulting from changes in the rate and isotopic
136
composition of soil CO2 production throughout diel cycles, would aid the interpretation of fieldderived surface CO2 flux measurements. Coupling Nickerson and Risk’s (2009c) and Risk and
Kellman’s (2008) models that incorporate the diffusivity of the individual 12CO2 and 13CO2
isotopologues with physicochemical models such as those used in Chapter 3 (and also by Ma et al.
(2013)) to describe the effects of temperature change on CO2 dissolution and exsolution would
provide a powerful means for predicting the rate and isotopic composition of abiotic CO2 production
in desert ecosystems. Temperature variation could easily be driven by mechanistic models simulating
heat transport within the soil profile; the models could be parameterised by data from this study.
6.4.3 Field soil CO2 flux experimental design
The data presented in this study highlight the need for high-resolution measurements of both surface
CO2 fluxes and subsurface CO2 concentration gradients. However, this study has demonstrated that
static chambers are not a reliable method for determining surface CO2 flux rates at biological
hotspots. Automated infra-red gas analysers, which have been used in previous studies of Dry Valley
soil CO2 flux rates (and can be adapted so as to provide greater sensitivity to the low flux rates typical
of Dry Valley soils (e.g. Parsons et al., 2004; Ball et al., 2009)), can determine surface CO2 flux rates
over considerably shorter measurement periods (up to 200 s; Parsons et al., 2004; Elberling et al.,
2006). A shorter measurement period would likely remove some of the high-frequency temporal
variability measured in static chambers at Site A, and the automated sampling mechanism would
enable a greater number of measurements to be made, thus providing a better opportunity to assess
fine-to-medium scale spatial variability.
However, measurements of surface CO2 flux rates alone, regardless of whether they coincide with a
period of steady-state subsurface CO2 concentrations, cannot provide any estimate as to the relative
contributions from biotic and abiotic sources. In the absence of C isotopic data, a surface CO2 flux
rate measured during a period of steady-state only represents the net rate of soil CO2 production,
which may comprise biotic and abiotic components. It is only when the δ13CΔCO2 of the net flux and
the δ13CCO2 of the biotic and abiotic components of that flux are known that the contributions of each
component can be quantified.
Soil CO2 fluxes as a biometric of change
Soil CO2 fluxes have considerable potential as a biometric that can be used to identify changes in Dry
Valley ecosystems. However, in order to provide a valuable, quantitative metric of change, surface
CO2 fluxes must be able to be partitioned into their biotic and abiotic components. At the very least,
this requires detailed measurements of subsurface CO2 concentration and δ13CCO2 which are currently
not easily obtained. However, development of promising new technology such as the Iso-FD
(Nickerson et al., 2013) should enable this to be achieved, with carefully managed automated
137
sensors able to provide considerable insight into soil CO2 flux dynamics beyond the short austral
summer field season. In particular, records over well-defined timescales (24-h, weekly, monthly,
seasonal and annual) will enable integration of the cyclical abiotic effects, thus, in the absence of any
long-term low-level continuous abiotic flux (e.g. geological; Risk et al., 2013), revealing any net
biological effect. It is only once the biotic and abiotic contributions are constrained that soil CO2
fluxes will provide a valuable long-term record of, and tool for predicting ecosystem response to,
future environmental change.
138
References
Adams, B.J., Bardgett, R.D., Ayres, E., Wall, D.H., Aislabie, J., Bamforth, S., Bargagli, R., Cary, C.,
Cavacini, P., Connell, L., Convey, P., Fell, J.W., Frati, F., Hogg, I.D., Newsham, K.K., O'Donnell,
A., Russell, N., Seppelt, R.D., Stevens, M.I., 2006. Diversity and distribution of Victoria Land
biota. Soil Biology & Biochemistry, 38, 3003-3018.
Amundson, R., Stern, L., Baisden, T., Wang, Y., 1998. The isotopic composition of soil and soilrespired CO2. Geoderma, 82, 83-114.
Anderson, R.S., Anderson, S.P., 2010. Geomorphology; the mechanics and chemistry of landscapes.
Cambridge University Press, Cambridge, 637 pp.
Ball, B.A., Virginia, R.A., Barrett, J.E., Parsons, A.N., Wall, D.H., 2009. Interactions between physical
and biotic factors influence CO2 flux in Antarctic dry valley soils. Soil Biology & Biochemistry,
41, 1510-1517.
Ball, B.A., Barrett, J.E., Gooseff, M.N., Virginia, R.A., Wall, D.H., 2011. Implications of meltwater pulse
events for soil biology and biogeochemical cycling in a polar desert. Polar Research, 30,
14555. doi:10.3402/polar.v30i0.14555
Bao, H., Marchant, D.R., 2006. Quantifying sulfate components and their variations in soils of the
McMurdo Dry Valleys, Antarctica. Journal of Geophysical Research, 111, D16301.
Bao, H., Barnes, J.D., Sharp, Z.D., Marchant, D.R., 2008. Two chloride sources in soils of the McMurdo
Dry Valleys, Antarctica. Journal of Geophysical Research-Atmospheres, 113.
doi:10.1029/2007jd008703
Barrett, J.E., Virginia, R.A., Parsons, A.N., Wall, D.H., 2005. Potential soil organic matter turnover in
Taylor Valley, Antarctica. Arctic, Antarctic, and Alpine Research, 37, 108-117.
Barrett, J.E., Virginia, R.A., Parsons, A.N., Wall, D.H., 2006a. Soil carbon turnover in the McMurdo Dry
Valleys, Antarctica. Soil Biology & Biochemistry, 38, 3065-3082.
Barrett, J.E., Virginia, R.A., Wall, D.H., Adams, B.J., 2008a. Decline in a dominant invertebrate species
contributes to altered carbon cycling in a low-diversity soil ecosystem. Global Change
Biology, 14, 1734-1744.
Barrett, J.E., Virginia, R.A., Wall, D.H., Parsons, A.N., Powers, L.E., Burkins, M.B., 2004. Variation in
biogeochemistry and soil biodiversity across spatial scales in a polar desert ecosystem.
Ecology, 85, 3105-3118.
Barrett, J.E., Virginia, R.A., Wall, D.H., Cary, S.C., Adams, B.J., Hacker, A.L., Aislabie, J.M., 2006b. Covariation in soil biodiversity and biogeochemistry in northern and southern Victoria Land,
Antarctica. Antarctic Science, 18, 535-548. doi:10.1017/s0954102006000587
Barrett, J.E., Virginia, R.A., Wall, D.H., Doran, P.T., Fountain, A.G., Welch, K.A., Lyons, W.B., 2008b.
Persistent effects of a discrete warming event on a polar desert ecosystem. Global Change
Biology, 14, 1-13. doi:10.1111/j.1365-2486.2008.01641.x
Barrett, J.E., Virginia, R.A., Hopkins, D.W., Aislabie, J., Bargagli, R., Bockheim, J.G., Campbell, I.B.,
Lyons, W.B., Moorhead, D.L., Nkem, J.N., Sletten, R.S., Steltzer, H., Wall, D.H., Wallenstein,
M.D., 2006c. Terrestrial ecosystem processes of Victoria Land, Antarctica. Soil Biology &
Biochemistry, 38, 3019-3034.
Barrett, P.J., Froggatt, P.C., 1978. Densities, porosities, and seismic velocities of some rocks from
Victoria Land, Antarctica. N.Z. Journal of Geology and Geophysics, 21, 175-187.
Bockheim, J.G., 1980. Properties and classification of some desert soils in coarse-textured glacial drift
in the Arctic and Antarctic. Geoderma, 24, 45-69.
Bockheim, J.G., 1990. Soil development rates in the Transantarctic Mountains. Geoderma, 47, 59-77.
139
Bockheim, J.G., 1997. Properties and classification of cold desert soils from Antarctica. Soil Science
Society of America Journal, 61, 224-231.
Bockheim, J.G., 2002. Landform and soil development in the McMurdo Dry Valleys, Antarctica: a
regional synthesis. Arctic, Antarctic, and Alpine Research, 34, 308-317.
Bockheim, J.G., 2010. Evolution of desert pavements and the vesicular layer in soils of the
Transantarctic Mountains. Geomorphology, 118, 433-443.
doi:10.1016/j.geomorph.2010.02.012
Bockheim, J.G., McLeod, M., 2008. Soil distribution in the McMurdo Dry Valleys, Antarctica.
Geoderma, 144, 43-49. doi:10.1016/j.geoderma.2007.10.015
Bockheim, J.G., Prentice, M.L., McLeod, M., 2008. Distribution of glacial deposits, soils, and
permafrost in Taylor Valley, Antarctica. Arctic, Antarctic, and Alpine Research, 40, 279-286.
doi:10.1657/1523-0430(06-057)[bockheim]2.0.co;2
Boström, B., Comstedt, D., Ekblad, A., 2007. Isotope fractionation and 13C enrichment in soil profiles
during the decomposition of soil organic matter. Oecologia, 153, 89-98.
Bowling, D.R., Massman, W.J., 2011. Persistent wind-induced enhancement of diffusive CO2 transport
in a mountain forest snowpack. Journal of Geophysical Research, 116, G04006.
Brummell, M.E., Farrell, R.E., Hardy, S.P., Siciliano, S.D., 2014. Greenhouse gas production and
consumption in High Arctic deserts. Soil Biology and Biochemistry, 68, 158-165.
doi:http://dx.doi.org/10.1016/j.soilbio.2013.09.034
Büdel, B., Bendix, J., Bicker, F.R., Green, T.G.A., 2008. Dewfall as a water source frequently activates
the endolithic cyanobacterial communities in the granites of Taylor Valley, Antarctica. Journal
of Phycology, 44, 1415-1424.
Burkins, M.B., Virginia, R.A., Wall, D.H., 2001. Organic carbon cycling in Taylor Valley, Antarctica:
quantifying soil reservoirs and soil respiration. Global Change Biology, 7, 113-125.
Burkins, M.B., Virginia, R.A., Chamberlain, C.P., Wall, D.H., 2000. Origin and Distribution of Soil
Organic Matter in Taylor Valley, Antarctica. Ecology, 81, 2377-2391.
Campbell, I.B., Claridge, G.G.C., 1975. Morphology and age relationships of Antarctic Soils. In:
Suggate, R.P., Cresswell, M.M. (Eds.), Quaternary Studies. The Royal Society of New Zealand,
Wellington, pp. 83-88.
Campbell, I.B., Claridge, G.G.C., 1982. The influence of moisture on the development of soils of the
cold deserts of Antarctica. Geoderma, 28, 221-238.
Campbell, I.B., Claridge, G.G.C., 2006. Permafrost properties, patterns and processes in the
Transantarctic Mountains region. Permafrost and Periglacial Processes, 17, 215-232.
doi:10.1002/ppp.557
Campbell, I.B., Claridge, G.G.C., Balks, M.R., Campbell, D.I., 1997. Moisture content in soils of the
McMurdo Sound and Dry Valley region of Antarctica. In: Lyons, W.B., Howard-Williams, C.,
Hawes, I. (Eds.), Ecosystem Processes in Antarctic Ice-Free Landscapes. A.A. Balkema,
Rotterdam, pp. 61-76.
Campbell, I.B., Claridge, G.G.C., Campbell, D.A., Balks, M.R., 1998. Permafrost properties in the
McMurdo Sound−Dry Valley region of Antarctica. In: Lewkowicz, A.G., Allard, M. (Eds.),
Proceedings of the Seventh International Conference on Permafrost, 121-126.
Cary, S.C., McDonald, I.R., Barrett, J.E., Cowan, D.A., 2010. On the rocks: the microbiology of
Antarctic Dry Valley Soils. Nature Reviews Microbiology, 8, 129-138.
Cerling, T.E., 1984. The stable isotopic composition of modern soil carbonate and its relationship to
climate. Earth and Planetary Science Letters, 71, 229-240.
140
Cerling, T.E., Solomon, D.K., Quade, J., Bowman, J.R., 1991. On the isotopic composition of carbon in
soil carbon dioxide. Geochimica et Cosmochimica Acta, 55, 3403-3405.
Claridge, G.G.C., 1965. The clay mineralogy and chemistry of some soils from the Ross Dependency,
Antarctica. New Zealand Journal of Geology and Geophysics, 8, 186-220.
Claridge, G.G.C., Campbell, I.B., 1977. The salts in Antarctic soils, their distribution and relationship to
soil processes. Soil Science, 123, 377-384.
Clarke, G.K.C., Waddington, E.D., 1991. A three-dimensional theory of wind pumping. Journal of
Glaciology, 37, 89-96.
Clow, G.D., McKay, C.P., Simmons, G.M., (Jr.), Wharton, R.A., (Jr.), 1988. Climatological observations
and predicted sublimation rates at Lake Hoare, Antarctica. Journal of Climate, 1, 715-728.
Colbeck, S.C., 1989. Air movement in snow due to windpumping. Journal of Glaciology, 35, 209-213.
Conen, F., Smith, K.A., 2000. An explanation of linear increases in gas concentration under closed
chambers used to measure gas exchange between soil and the atmosphere. European
Journal of Soil Science, 51, 111-117.
Conovitz, P.A., McKnight, D.M., MacDonald, L.H., Fountain, A.G., House, H.R., 1998. Hydrologic
processes influencing streamflow variation in Fryxell Basin, Antarctica. In: Priscu, J.C. (Ed.),
Ecosystem dynamics in a polar desert: The McMurdo Dry Valleys, Antarctica, vol. 72.
American Geophysical Union, Washington, D.C., pp. 93-108.
Craig, H., 1953. The geochemistry of the stable carbon isotopes. Geochimica et Cosmochimica Acta,
3, 53-92.
Creelman, C., Nickerson, N., Risk, D., 2013. Quantifying lateral diffusion error in soil carbon dioxide
respiration estimates using numerical modeling. Soil Science Society of America Journal, 77,
699-708. doi:10.2136/sssaj2012.0352
Dana, G.L., Wharton, R.A., (Jr.), Dubayah, R., 1998. Solar radiation in the McMurdo Dry Valleys,
Antarctica. In: Priscu, J.C. (Ed.), Ecosystem dynamics in a polar desert: The McMurdo Dry
Valleys, Antarctica. American Geophysical Union, Washington, D.C., pp. 39-64.
Davidson, E.A., Savage, K., Verchot, L.V., Navarro, R., 2002. Minimizing artifacts and biases in
chamber-based measurements of soil respiration. Agricultural and Forest Meteorology, 113,
21-37.
Denton, G.H., Bockheim, J.G., Wilson, S.C., Stuiver, M., 1989. Late Wisconsin and early Holocene
glacial history, inner Ross Embayment, Antarctica. Quaternary Research, 31, 151-182.
doi:10.1016/0033-5894(89)90004-5
Doran, P.T., McKay, C.P., Clow, G.D., Dana, G.L., Fountain, A.G., Nylen, T., Lyons, W.B., 2002a. Valley
floor climate observations from the McMurdo dry valleys, Antarctica, 1986-2000. Journal of
Geophysical Research-Atmospheres, 107
Doran, P.T., McKay, C.P., Fountain, A.G., Nylen, T., McKnight, D.M., Jaros, C., Barrett, J.E., 2008.
Hydrologic response to extreme warm and cold summers in the McMurdo Dry Valleys, East
Antarctica. Antarctic Science, 20, 499-509. doi:10.1017/s0954102008001272
Doran, P.T., Priscu, J.C., Lyons, W.B., Walsh, J.E., Fountain, A.G., McKnight, D.M., Moorhead, D.L.,
Virginia, R.A., Wall, D.H., Clow, G.D., Fritsen, C.H., McKay, C.P., Parsons, A.N., 2002b.
Antarctic climate cooling and terrestrial ecosystem response. Nature, 415, 517-520.
Dörr, H., Münnich, K.O., 1980. Carbon-14 and carbon-13 in soil CO2. Radiocarbon, 22, 909-918.
Ekblad, A., Högberg, P., 2000. Analysis of δ13C of CO2 distinguishes between microbial respiration of
added C4-sucrose and other soil respiration in a C3-ecosystem. Plant and Soil, 219, 197-209.
141
Elberling, B., Gregorich, E.G., Hopkins, D.W., Sparrow, A.D., Novis, P., Greenfield, L.G., 2006.
Distribution and dynamics of soil organic matter in an Antarctic dry valley. Soil Biology &
Biochemistry, 38, 3095-3106.
Elberling, B., Greenfield, L.G., Gregorich, E.G., Hopkins, D.W., Novis, P., Sparrow, A.D., 2013.
Comments on "Abiotic processes dominate CO2 flux in Antarctic soils" by Shanhun et al. Soil
Biology & Biochemistry 53, 99-111 (2012). Soil Biology & Biochemistry, In Press.
doi:http://dx.doi.org/10.1016/j.soilbio.2013.10.046
Fang, C., Moncrieff, J.B., 1999. A model for soil CO2 production and transport 1: Model development.
Agricultural and Forest Meteorology, 95, 225-236.
Fang, C., Moncrieff, J.B., 2001. The dependence of soil CO2 efflux on temperature. Soil Biology &
Biochemistry, 33, 155-165.
Foley, K.K., 2005. Pedogenic carbonate distribution within glacial till in Taylor Valley, southern
Victoria Land, Antarctica, vol. Master of Science. The Ohio State University, pp. 100.
Foley, K.K., Lyons, W.B., Barrett, J.E., Virginia, R.A., 2006. Pedogenic carbonate distribution within
glacial till in Taylor Valley, Southern Victoria Land, Antarctica. In: Alonso-Zarza, A.M., Tanner,
L.H. (Eds.), Paleoenvironmental Record and Applications of Calcretes and Palustrine
Carbonates: Geological Society of America Special Paper 416. Geological Society of America,
pp. 89-103.
Fountain, A.G., Nylen, T.H., Monaghan, A., Basagic, H.J., Bromwich, D., 2010. Snow in the McMurdo
Dry Valleys, Antarctica. International Journal of Climatology, 30, 633-642.
doi:10.1002/joc.1933
Fountain, A.G., Lyons, W.B., Burkins, M.B., Dana, G.L., Doran, P.T., Lewis, K.J., McKnight, D.M.,
Moorhead, D.L., Parsons, A.N., Priscu, J.C., Wall, D.H., Wharton, R.A., (Jr.), Virginia, R.A.,
1999. Physical controls on the Taylor Valley ecosystem, Antarctica. BioScience, 49, 961-971.
Freckman, D.W., Virginia, R.A., 1997. Low-diversity Antarctic soil nematode communities:
Distribution and response to disturbance. Ecology, 78, 363-369.
Fry, B., 2006. Stable Isotope Ecology. Springer, New York
Fukuda, H., 1955. Air and vapor movement in soil due to wind gustiness. Soil Science, 79, 249-256.
Gamnitzer, U., Moyes, A.B., Bowling, D.R., Schnyder, H., 2011. Measuring and modelling the isotopic
composition of soil respiration: insights from a grassland tracer experiment. Biogeosciences,
8, 1333-1350.
Gregorich, E.G., Hopkins, D.W., Elberling, B., Sparrow, A.D., Novis, P., Greenfield, L.G., Rochette, P.,
2006. Emission of CO2, CH4 and N2O from lakeshore soils in an Antarctic dry valley. Soil
Biology & Biochemistry, 38, 3120-3129.
Hall, B.L., Denton, G.H., 2000. Radiocarbon Chronology of Ross Sea Drift, Eastern Taylor Valley,
Antarctica: Evidence for a Grounded Ice Sheet in the Ross Sea at the Last Glacial Maximum.
Geografiska Annaler: Series A, Physical Geography, 82, 305-336. doi:10.1111/j.04353676.2000.00127.x
Hall, B.L., Denton, G.H., Hendy, C.H., 2000. Evidence from Taylor Valley for a grounded ice sheet in
the Ross Sea, Antarctica. Geografiska Annaler: Series A, Physical Geography, 82, 275-303.
doi:10.1111/j.0435-3676.2000.00126.x
Harris, D., Horwath, W.R., van Kessel, C., 2001. Acid fumigation of soils to remove carbonates prior to
total organic carbon or carbon-13 isotopic analysis. Soil Science Society of America Journal,
65, 1853-1856.
Healy, R.W., Striegl, R.G., Russell, T.F., Hutchinson, G.L., Livingston, G.P., 1996. Numerical evaluation
of static-chamber measurements of soil−atmosphere gas exchange: identification of physical
processes. Soil Science Society of America Journal, 60, 740-747.
142
Hendy, C.H., 2000. Late Quaternary lakes in the McMurdo Sound region of Antarctica. Geografiska
Annaler: Series A, Physical Geography, 82, 411-432. doi:10.1111/j.0435-3676.2000.00131.x
Hendy, C.H., Healy, T.R., Rayner, E.M., Shaw, J., Wilson, A.T., 1979. Late Pleistocene glacial
chronology of the Taylor Valley, Antarctica, and the global climate. Quaternary Research, 11,
172-184.
Higgins, S., Denton, G.H., Hendy, C.H., 2000a. Glacial Geomorphology of Bonney Drift, Taylor Valley,
Antarctica. Geografiska Annaler: Series A, Physical Geography, 82, 365-389.
doi:10.1111/j.0435-3676.2000.00129.x
Higgins, S.M., Hendy, C.H., Denton, G.H., 2000b. Geochronology of Bonney Drift, Taylor Valley,
Antarctica: Evidence for Interglacial Expansions of Taylor Glacier. Geografiska Annaler: Series
A, Physical Geography, 82, 391-409. doi:10.1111/j.0435-3676.2000.00130.x
Hoefs, J., 2009. Stable Isotope Geochemistry. Springer-Verlag, Berlin
Hogg, I.D., Cary, S.C., Convey, P., Newsham, K.K., O'Donnell, A.G., Adams, B.J., Aislabie, J., Frati, F.,
Stevens, M.I., Wall, D.H., 2006. Biotic interactions in Antarctic terrestrial ecosystems: Are
they a factor? Soil Biology & Biochemistry, 38, 3035-3040.
Hopkins, D.W., Sparrow, A.D., Elberling, B., Gregorich, E.G., Novis, P.M., Greenfield, L.G., Tilston, E.L.,
2006. Carbon, nitrogen and temperature controls on microbial activity in soils from an
Antarctic dry valley. Soil Biology & Biochemistry, 38, 3130-3140.
Hopkins, D.W., Sparrow, A.D., Shillam, L.L., English, L.C., Dennis, P.G., Novis, P., Elberling, B.,
Gregorich, E.G., Greenfield, L.G., 2008. Enzymatic activities and microbial communities in an
Antarctic dry valley soil: Responses to C and N supplementation. Soil Biology & Biochemistry,
40, 2130-2136.
Hopkins, D.W., Sparrow, A.D., Gregorich, E.G., Elberling, B., Novis, P., Fraser, F., Scrimgeour, C.,
Dennis, P.G., Meier-Augenstein, W., Greenfield, L.G., 2009. Isotopic evidence for the
provenance and turnover of organic carbon by soil microorganisms in the Antarctic dry
valleys. Environmental Microbiology, 11, 597-608. doi:10.1111/j.1462-2920.2008.01830.x
Keeling, C.D., Piper, S.C., Bacastow, R.B., Wahlen, M., Whorf, T.P., Heimann, M., Meijer, H.A., 2001.
Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978
to 2000. I. Global aspects, SIO Reference Series, No. 01-06. Scripps Institution of
Oceanography, San Diego, pp. 88. http://scrippsco82.ucsd.edu/data/atmospheric_co82.html.
Kendall, C., Caldwell, E.A., 1998. Fundamentals of Isotope Geochemistry. In: Kendall, C., McDonnell,
J.J. (Eds.), Isotope Tracers in Catchment Hydrology. Elsevier Science B.V., Amsterdam, pp. 5186.
Keys, J.R., 1980. Air temperature, wind, precipitation and atmospheric humidity in the McMurdo
Region. Department of Geology Publication No. 17, Victoria University of Wellington,
Wellington, NZ, 57 pp.
Keys, J.R., Williams, K., 1981. Origin of crystalline, cold desert salts in the McMurdo region,
Antarctica. Geochimica Et Cosmochimica Acta, 45, 2299-2309.
Kirschbaum, M.U.F., 1995. The temperature dependence of soil organic matter decomposition, and
the effect of global warming on soil organic C storage. Soil Biology & Biochemistry, 27, 753760.
Knohl, A., Werner, R.A., Geilmann, H., Brand, W.A., 2004. Kel-FTM discs improve storage time of
canopy air samples in 10-mL vials for CO2-δ13C analysis. Rapid Communications in Mass
Spectrometry, 18, 1663-1665.
Lee, C.K., Barbier, B.A., Bottos, E.M., McDonald, I.R., Cary, S.C., 2012. The inter-valley soil
comparative survey: the ecology of Dry Valley edaphic microbial communities. The ISME
Journal, 6, 1046-1057. doi:10.1038/ismej.2011.170
143
Levy, J., 2013. How big are the McMurdo Dry Valleys? Estimating ice-free area using Landsat image
data. Antarctic Science, 25, 119-120. doi:10.1017/S0954102012000727
Levy, J.S., Fountain, A.G., Gooseff, M.N., Welch, K.A., Lyons, W.B., 2011. Water tracks and permafrost
in Taylor Valley, Antarctica: Extensive and shallow groundwater connectivity in a cold desert
ecosystem. GSA Bulletin, 123, 2295-2311.
Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Functional Ecology, 8,
315-323.
Lyons, W.B., Welch, K.A., Carey, A.E., Doran, P.T., Wall, D.H., Virginia, R.A., Fountain, A.G., Csathó,
B.M., Tremper, C.M., 2005. Groundwater seeps in Taylor Valley Antarctica: an example of a
subsurface melt event. Annals of Glaciology, 40, 200-206.
Ma, J., Wang, Z.Y., Stevenson, B.A., Zheng, X.J., Li, Y., 2013. An inorganic CO2 diffusion and dissolution
process explains negative CO2 fluxes in saline/alkaline soils. Scientific Reports, 3, 7.
doi:10.1038/srep02025
Magalhães, C., Stevens, M.I., Cary, S.C., Ball, B.A., Storey, B.C., Wall, D.H., Türk, R., Ruprecht, U.,
2012. At limits of life: Multidisciplinary insights reveal environmental constraints on biotic
diversity in continental Antarctica. Plos One, 7, e44578. doi:10.1371/journal.pone.0044578
Marrero, T.R., Mason, E.A., 1972. Gaseous Diffusion Coefficients. Journal of Physical and Chemical
Reference Data, 1, 1-118.
Massman, W.J., 2006. Advective transport of CO2 in permeable media induced by atmospheric
pressure fluctuations: 1. An analytical model. Journal of Geophysical Research, 111, G03004.
Massman, W.J., Sommerfeld, R.A., Mosier, A.R., Zeller, K.F., Hehn, T.J., Rochelle, S.G., 1997. A model
investigation of turbulence-driven pressure-pumping effects on the rate of diffusion of CO2,
N2O, and CH4 through layered snowpacks. Journal of Geophysical Research, 102, 18,851818,863.
McCarthy, K.A., Johnson, R.L., 1995. Measurement of trichloroethylene diffusion as a function of
moisture content in sections of gravity-drained soil columns. Journal of Environmental
Quality, 24, 49-55.
McCraw, J.D., 1967. Soils of Taylor Dry Valley, Victoria Land, Antarctica, with notes on soils from
other localities in Victoria Land. New Zealand Journal of Geology and Geophysics, 10, 498539.
McKay, C.P., Mellon, M.T., Friedmann, E.I., 1998. Soil temperatures and stability of ice-cemented
ground in the McMurdo Dry Valleys, Antarctica. Antarctic Science, 10, 31-38.
McKnight, D.M., Niyogi, D.K., Alger, A.S., Bomblies, A., Conovitz, P.A., Tate, C.M., 1999. Dry valley
streams in Antarctica: Ecosystems waiting for water. Bioscience, 49, 985-995.
Mergelov, N.S., Goryachkin, S.V., Shorkunov, I.G., Zazovskaya, E.P., Cherkinsky, A.E., 2012. Endolithic
pedogenesis and rock varnish on massive crystalline rocks in East Antarctica. Eurasian Soil
Science, 45, 901-917.
Michalski, G., Bockheim, J.G., Kendall, C., Thiemens, M., 2005. Isotopic composition of Antarctic Dry
Valley nitrate: Implications for NOy sources and cycling in Antarctica. Geophysical Research
Letters, 32, L13817.
Midwood, A.J., Millard, P., 2011. Challenges in measuring the δ13C of the soil surface CO2 efflux. Rapid
Communications in Mass Spectrometry, 25, 232-242. doi:10.1002/rcm.4857
Moldrup, P., Olesen, T., Yoshikawa, S., Komatsu, T., Rolston, D.E., 2004. Three-porosity model for
predicting the gas diffusion coefficient in undisturbed soil. Soil Science Society of America
Journal, 68, 750-759.
144
Mook, W.G., Bommerson, J.C., Staverman, W.H., 1974. Carbon isotope fractionation between
dissolved bicarbonate and gaseous carbon dioxide. Earth and Planetary Science Letters, 22,
169-176. doi:10.1016/0012-821x(74)90078-8
Mook, W.G. (Ed.). 2000. Environmental isotopes in the hydrological cycle: Principles and applications
(Volume 1). Paris: UNESCO.
Moorhead, D.L., 2007. Mesoscale dynamics of ephemeral wetlands in the Antarctic Dry Valleys:
Implications to production and distribution of organic matter. Ecosystems, 10, 86-94.
Moorhead, D.L., Barrett, J.E., Virginia, R.A., Wall, D.H., Porazinska, D., 2003. Organic matter and soil
biota of upland wetlands in Taylor Valley, Antarctica. Polar Biology, 26, 567-576.
Moyes, A.B., Gaines, S.J., Siegwolf, R.T.W., Bowling, D.R., 2010. Diffusive fractionation complicates
isotopic partitioning of autotrophic and heterotrophic sources of soil respiration. Plant, Cell
and Environment, 33, 1804-1819.
Nickerson, N., Risk, D., 2009a. A numerical evaluation of chamber methodologies used in measuring
the δ13C of soil respiration. Rapid Communications in Mass Spectrometry, 23, 2802-2810.
doi:10.1002/rcm.4189
Nickerson, N., Risk, D., 2009b. Keeling plots are non-linear in non-steady state diffusive
environments. Geophysical Research Letters, 36, L08401. doi:10.1029/2008GL036945
Nickerson, N., Risk, D., 2009c. Physical controls on the isotopic composition of soil-respired CO2.
Journal of Geophysical Research, 114, G01013. doi:10.1029/2008JG000766
Nickerson, N., Egan, J., Risk, D., 2013. Iso-FD: A novel method for measuring the isotopic signature of
surface flux. Soil Biology & Biochemistry, 62, 99-106.
Nielsen, U.N., Wall, D.H., Adams, B.J., Virginia, R.A., Ball, B.A., Gooseff, M.N., McKnight, D.M., 2012.
The ecology of pulse events: insights from an extreme climatic event in a polar desert
ecosystem. Ecosphere, 3, Article 17. doi:http://dx.doi.org/10.1890/ES11-00325.1
Nylen, T.H., Fountain, A.G., Doran, P.T., 2004. Climatology of katabatic winds in the McMurdo dry
valleys, southern Victoria Land, Antarctica. Journal of Geophysical Research-Atmospheres,
109. doi:10.1029/2003jd003937
O'Leary, M.H., 1981. Carbon isotope fractionation in plants. Phytochemistry, 20, 553-567.
O'Leary, M.H., 1988. Carbon isotopes in photosynthesis. Bioscience, 38, 328-336.
Park, R., Epstein, S., 1960. Carbon isotope fractionation during photosynthesis. Geochimica et
Cosmochimica Acta, 21, 110-126.
Parsons, A.N., Barrett, J.E., Wall, D.H., Virginia, R.A., 2004. Soil carbon dioxide flux in Antarctic Dry
Valley ecosystems. Ecosystems, 7, 286-295.
Pastor, J., Bockheim, J.G., 1980. Soil development on moraines of Taylor Glacier, lower Taylor Valley,
Antarctica. Soil Science Society of America Journal, 44, 341-348.
Poage, M.A., Barrett, J.E., Virginia, R.A., Wall, D.H., 2008. The influence of soil geochemistry on
nematode distribution, McMurdo Dry Valleys, Antarctica. Arctic, Antarctic, and Alpine
Research, 40, 119-128.
Pumpanen, J., Kolari, P., Ilvesniemi, H., Minkkinen, K., Vesala, T., Niinistö, S., Lohila, A., Larmola, T.,
Morero, M., Pihlatie, M., Janssens, I., Yuste, J.C., Grünzweig, J.M., Reth, S., Subke, J.-A.,
Savage, K., Kutsch, W., Østreng, G., Ziegler, W., Anthoni, P., Lindroth, A., Hari, P., 2004.
Comparison of different chamber techniques for measuring soil CO2 efflux. Agricultural and
Forest Meteorology, 123, 159-176. doi:10.1016/j.agrformet.2003.12.001
Rayment, M.B., 2000. Closed chamber systems underestimate soil CO2 efflux. European Journal of
Soil Science, 51, 107-110.
145
Risk, D., Kellman, L., 2008. Isotopic fractionation in non-equilibrium diffusive environments.
Geophysical Research Letters, 35, L02403. doi:10.1029/2007/GL032374
Risk, D., Lee, C.K., MacIntyre, C., Cary, S.C., 2013. First year-round record of Antarctic Dry Valley soil
CO2 flux. Soil Biology & Biochemistry, 66, 193-196.
Rochette, P., Bertrand, N., 2003. Soil air sample storage and handling using polypropylene syringes
and glass vials. Canadian Journal of Soil Science, 631-637.
Rochette, P., Flanagan, L.B., Gregorich, E.G., 1999. Separating soil respiration into plant and soil
components using analyses of the natural abundance of carbon-13. Soil Science Society of
America Journal, 63, 1207-1213. doi:10.2136/sssaj1999.6351207x
Smith, J.L., Barrett, J.E., Tusnady, G., Rejto, L., Cary, S.C., 2010. Resolving environmental drivers of
microbial community structure in Antarctic soils. Antarctic Science, 22, 673-680.
doi:10.1017/s0954102010000763
Sparrow, A.D., Gregorich, E.G., Hopkins, D.W., Novis, P., Elberling, B., Greenfield, L.G., 2011. Resource
limitations on soil microbial activity in an Antarctic dry valley. Soil Science Society of America
Journal, 75, 2188-2197.
Staff, U.S.S.L., 1954. Diagnosis and Improvement of Saline and Alkali Soils. United States Department
of Agriculture, Washington, D.C.
Stomeo, F., Makhalanyane, T.P., Valverde, A., Pointing, S.B., Stevens, M.I., Cary, S.C., Tuffin, M.I.,
Cowan, D.A., 2012. Abiotic factors influence microbial diversity in permanently cold soil
horizons of a maritime-associated Antarctic Dry Valley. FEMS Microbiology Ecology, 82, 326340.
Stumm, W., Morgan, J.J., 1996. Aquatic chemistry: chemical equilibria and rates in natural waters,
3rd ed. Wiley Interscience, New York
Subke, J.-A., Bahn, M., 2010. On the 'temperature sensitivity' of soil respiration: Can we use the
immeasurable to predict the unknown? Soil Biology & Biochemistry, 42, 1653-1656.
Takle, E.S., Massman, W.J., Brandle, J.R., Schmidt, R.A., Zhou, X., Litvina, I.V., Garcia, R., Doyle, G.,
Rice, C.W., 2004. Influence of high-frequency ambient pressure pumping on carbon dioxide
efflux from soil. Agricultural and Forest Meteorology, 124, 193-206.
Thode, H.G., Shima, M., Rees, C.F., Krishnamurty, K.V., 1965. Carbon-13 isotope effects in systems
containing carbon dioxide, bicarbonate, carbonate and metal ions. Canadian Journal of
Chemistry, 43, 582-595.
Thompson, D.C., Craig, R.M.F., Bromley, A.M., 1971. Climate and surface heat balance in an Antarctic
Dry Valley. New Zealand Journal of Science, 14, 245-251.
Toner, J.D., Sletten, R.S., Prentice, M.L., 2013. Soluble salt accumulations in Taylor Valley, Antarctica:
Implications for paleolakes and Ross Sea Ice Sheet dynamics. Journal of Geophysical
Research: Earth Surface, 118, 198-215. doi:10.1029/2012JF002467
Treonis, A.M., Wall, D.H., 2005. Soil nematodes and desiccation survival in the extreme arid
environment of the Antarctic Dry Valleys. Integrative and Comparative Biology, 45, 741-750.
Treonis, A.M., Wall, D.H., Virginia, R.A., 1999. Invertebrate biodiversity in Antarctic dry valley soils
and sediments. Ecosystems, 2, 482-492.
Treonis, A.M., Wall, D.H., Virginia, R.A., 2000. The use of anhydrobiosis by soil nematodes in the
Antarctic Dry Valleys. Functional Ecology, 14, 460-467. doi:10.1046/j.13652435.2000.00442.x
Treonis, A.M., Wall, D.H., Virginia, R.A., 2002. Field and microcosm studies of decomposition and soil
biota in a cold desert soil. Ecosystems, 5, 159-170.
146
Van Horn, D.J., Van Horn, M.L., Barrett, J.E., Gooseff, M.N., Altrichter, A.E., Geyer, K.M., Zeglin, L.H.,
Takacs-Vesbach, C.D., 2013. Factors Controlling Soil Microbial Biomass and Bacterial Diversity
and Community Composition in a Cold Desert Ecosystem: Role of Geographic Scale. Plos One,
8, e66103. doi:10.1371/journal.pone.0066103
Vogel, J.C., Grootes, P.M., Mook, W.G., 1970. Isotope fractionation between gaseous and dissolved
carbon dioxide. Zeitschrift für Physik, 230, 225-238.
Wall, D.H., Virginia, R.A., 1999. Controls on soil biodiversity: insights from extreme environments.
Applied Soil Ecology, 13, 137-150.
Werth, M., Kuzyakov, Y., 2010. 13C fractionation at the root−microorganisms−soil interface: A review
and outlook for partitioning studies. Soil Biology & Biochemistry, 42, 1372-1384.
Wilch, T.I., Denton, G.H., Lux, D.R., McIntosh, W.C., 1993. Limited Pliocene glacier extent and surface
uplift in middle Taylor Valley, Antarctica. Geografiska Annaler, 75 A 331-351.
Zhu, B., Cheng, W., 2011. Constant and diurnally-varying temperature regimes lead to different
temperature sentivities of soil organic carbon decomposition. Soil Biology & Biochemistry,
43, 866-869.
147
148
Appendix A
Publications resulting from this thesis
149
Soil Biology & Biochemistry 53 (2012) 99e111
Contents lists available at SciVerse ScienceDirect
Soil Biology & Biochemistry
journal homepage: www.elsevier.com/locate/soilbio
Abiotic processes dominate CO2 fluxes in Antarctic soils
Fiona L. Shanhun*, Peter C. Almond, Tim J. Clough, Carol M.S. Smith
Department of Soil and Physical Sciences, PO Box 84, Lincoln University, Lincoln 7647, New Zealand
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 20 September 2011
Received in revised form
5 April 2012
Accepted 19 April 2012
Available online 18 May 2012
Ecosystems within the McMurdo Dry Valleys of Antarctica are highly sensitive to environmental change.
Increases in soil temperature and/or moisture content may dramatically change rates of soil respiration
and soil carbon (C) turnover. Present estimates of soil respiration rates and C turnover times are based on
surface carbon dioxide (CO2) fluxes and soil organic C content. However, the assumption that surface CO2
fluxes are purely biological in origin has not been rigorously tested. We use concentration and, for the
first time, the stable C isotopic composition of surface soil CO2 fluxes and subsurface CO2 profiles to: 1)
examine mechanisms of soil CO2 uptake and release, 2) identify the location of potential CO2 sources and
sinks within the soil profile, and 3) discriminate between biotic and abiotic contributions to CO2 fluxes in
soils of Taylor Valley. Surface CO2 fluxes and subsurface CO2 profiles confirm that these soils take up and
release CO2 on a daily basis (during the austral summer), associated with small changes in soil
temperature. Shifts in the C isotopic composition of soil CO2 are inconsistent with biological mechanisms
of CO2 production and consumption. Instead, the isotopic shifts can be accounted for by Henry’s Law
dissolution and exsolution of CO2 into a solution of high pH, driven by changes in soil temperature. Our
results constrain the biological component of soil CO2 fluxes in Taylor Valley to less than 25% (and likely
to be significantly less). This finding implies that previous measurements of surface soil CO2 fluxes are
overestimates of soil respiration, thus C turnover times calculated from them are underestimates.
Discriminating between biotic and abiotic contributions to CO2 fluxes in Antarctic dry valley soils is
essential if the effects of climate change on these sensitive ecosystems are to be accurately identified.
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
Soil respiration
d13CCO2
Organic C
Inorganic C
McMurdo Dry Valleys
Antarctic soils
1. Introduction
Predicting the impact of changes in climate conditions on
Antarctic ice-free areas necessitates a detailed understanding of the
rates and dynamics of processes operating in these extreme environments. Previous studies of soil respiration in Antarctic ecosystems
have been conducted in order to quantify the turnover time of the soil
organic carbon (C) pool (Burkins et al., 2001; Barrett et al., 2006a;
Elberling et al., 2006), and to provide a basis for comparing ecosystem
functioning between sites (Barrett et al., 2006b). Investigations of
Antarctic soil respiration rates have also provided insight into the
physical controls on soil carbon dioxide (CO2) flux (Parsons et al.,
2004) and the effects of climate variation on C cycling (Ball et al.,
2009) in dry valley ecosystems. Surface CO2 fluxes in the McMurdo
Dry Valleys generally range from 0.1 to 0.15 mmol CO2 m2 s1 (e.g.
Burkins et al., 2001; Parsons et al., 2004; Elberling et al., 2006; Ball
et al., 2009), although at sites with higher C and moisture contents,
* Corresponding author. Tel.: þ64 3 321 8246; fax: þ64 3 325 3607.
E-mail addresses: fiona.shanhun@lincoln.ac.nz (F.L. Shanhun), peter.almond@
lincoln.ac.nz
(P.C.
Almond),
timothy.clough@lincoln.ac.nz
(T.J.
Clough),
carol.smith@lincoln.ac.nz (C.M.S. Smith).
0038-0717/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soilbio.2012.04.027
rates can be as high as 0.78 mmol CO2 m2 s1 (Gregorich et al., 2006).
Previous studies have generally focused on relating soil respiration
rates to environmental variables such as soil temperature and
moisture content (Parsons et al., 2004; Elberling et al., 2006;
Gregorich et al., 2006) and observing responses to in situ (Burkins
et al., 2001; Ball et al., 2009) and laboratory manipulations (Burkins
et al., 2001; Treonis et al., 2002; Parsons et al., 2004; Barrett et al.,
2005, 2006a; Elberling et al., 2006; Hopkins et al., 2006, 2008, 2009).
Studies by Parsons et al. (2004) and Ball et al. (2009) found that
soil CO2 fluxes in the McMurdo Dry Valleys follow diel variations,
with soil temperature accounting for almost half of the variation in
CO2 flux (Ball et al., 2009). Both authors conclude that soil CO2 flux
is strongly influenced by abiotic factors, and suggest that dissolution of CO2 in soil water, consistent with Henry’s Law, is the likely
explanation for the diel variation observed. This is in contrast to
many other ecosystems, where temperature predominantly influences biotic factors.
An abiotic contribution to soil CO2 flux has potentially profound
implications for interpretation of CO2 fluxes in terms of soil respiration and C cycling in the McMurdo Dry Valleys. Present estimates
of turnover times based on in situ soil respiration rates are between
23 and 130 years (Burkins et al., 2001; Elberling et al., 2006). Any
100
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
abiotic contributions to CO2 fluxes would mean that soil respiration
rates overestimate the biological contributions, and turnover times
calculated from such fluxes would be underestimates.
In this paper, we present data from a study aimed to advance the
mechanistic understanding of the processes governing CO2 fluxes
in Antarctic soils. We use surface soil CO2 fluxes and subsurface soil
CO2 concentration profiles to characterise diel soil CO2 dynamics,
and identify potential CO2 sources and sinks within the soil profile.
We then test the hypothesis of Ball et al. (2009) and Parsons et al.
(2004) by: 1) comparing measured subsurface soil CO2 concentrations with values simulated by a model that incorporates Henry’s
Law dissolution of CO2, and 2) quantifying the biotic and abiotic
components of soil CO2 by utilising the contrasting relationships
between concentration and C isotopic composition of soil CO2
13
(d CCO2 ) characteristic of biotic and abiotic processes.
2. Materials and methods
2.1. Study area
The study was conducted during the 2008/09 austral summer at
two sites in Taylor Valley, in the McMurdo Dry Valleys, Antarctica.
The McMurdo Dry Valleys are the largest ice-free region in
Antarctica (w4800 km2) and comprise a polar desert landscape
dominated by coarse-textured soils formed in glacial drift, interspersed with alpine glaciers, ephemeral streams, and perennially
ice-covered lakes. Vascular vegetation is absent, and biota consists
of microorganisms and a limited number of invertebrate species
(Adams et al., 2006; Cary et al., 2010). Mean annual air temperature
in Taylor Valley ranges from 14.8 C to 23.1 C (Doran et al.,
2002), and annual water-equivalent precipitation is <50 mm,
with snow transported by katabatic winds supplying up to another
50 mm (water-equivalent) to the valley floor (Fountain et al., 2010).
Site A was on relatively young glacial drift (Ross Sea drift,
w9e28 ka; Hall and Denton, 2000), adjacent to a small lake
(w0.05 km2) situated 700 m east of Howard Glacier (77 39.5720 S,
163 07.3940 E). This site was selected to provide soils with relatively
low salt concentrations and higher inputs of organic C. During
summer, when a moat forms around the edge of the lake, algae are
driven to the lake margins by winds and are eventually blown
onshore, providing a local source of organic C. Soils at Site A are
weakly developed, strongly cryoturbated, and contain ice-cemented
permafrost within 70 cm of the soil surface (Typic Haploturbels;
Bockheim et al., 2008). At the time of sampling (late December
2008), soils at Site A were ice-cemented below 35 cm depth. Site B
was on an older moraine crest (Taylor III drift, 250e340 ka; Higgins
et al., 2000), approximately 1.5 km west of Howard Glacier (77
39.5400 S, 163 00.4840 E). Soils at Site B were expected to have higher
salt concentrations due to their age, and lower organic C levels as
there was no lake nearby to provide a local source of organic
material. Soils formed in Taylor III drift are mapped as Typic
Anhyorthels (Bockheim et al., 2008), which by definition should not
contain ice-cemented permafrost in the upper 70 cm of the profile.
However, soils at Site B had a similar morphology to those at Site A
(Typic Haploturbels), and were ice-cemented below 26 cm depth.
Both sites were characterised by patterned ground comprised of
flat-centred polygons measuring approximately 15 15 m.
2.2. Soil temperature and moisture content measurements
Soil temperature and volumetric moisture content at 5, 15 and
25 cm depth were measured using Hydra Probes (Hydra Probe II,
Stevens Water Monitoring Systems Inc., Portland, Oregon, USA). An
intermittent fault occurred in the probe installed at 5 cm depth at
Site A; data from this probe are only available for one of the CO2
sampling periods. Surface air temperature was measured 2 mm
above the soil surface (CS215-L, Campbell Scientific Inc., Logan, Utah,
USA). All data were recorded every 2 min, with 20 min averages
calculated and logged by a CR 1000 data logger (Campbell Scientific
Inc., Logan, Utah, USA).
2.3. Soil CO2 sampling
Soil CO2 sampling was replicated in two adjacent polygons at
each site. Within each polygon, five opaque tubes (iPLexÒ Novadrain uPVC access tubes e 15 cm diameter by 16 cm high; iPLEXÒ
DWV 136.150) were pressed into the soil surface to a depth of 6 cm.
An opaque uPVC lid (iPLEXÒ DWV 136C.150) with an internal
rubber O-ring, drilled to fit a pressure release bung and an injectable septum (Baxter Interlink Injection Site Ref #2N3399K), was
screwed on to the tubes immediately before surface CO2 flux
sampling. This created a chamber with a mean effective headspace
of 0.0018 m3. In order to determine surface soil CO2 fluxes, 20 ml
gas samples were taken from the headspace chambers immediately
after the chamber lids were fitted and vented, with three subsequent samples taken at 20 min intervals over a 1 h period. Surface
CO2 fluxes were calculated based on headspace CO2 concentrations
during the 1 h sampling period using the following equation:
nCO2 ¼ ½CO2 PV
RTA
(1)
where nCO2 is the number of moles of CO2 emitted per m2
(mmol m2), [CO2] is headspace CO2 concentration (mL L1), P is
atmospheric pressure (101325 Pa), V is the chamber volume (m3), R
is the ideal gas constant (8.314 J K1 mol1), T is air temperature (K),
and A is the area covered by the chamber (0.0177 m2). Surface CO2
flux rates (mmol m2 s1) were calculated by linear regression of
surface CO2 fluxes (mmol m2) against time over which samples were
collected (s). Linear regression was also used to determine the rate of
13
change of d CCO2 (& min1) in the headspace chamber samples.
Soil profile CO2 samples were taken from polyethylene subsurface sampling tubes (Bev-A-LineÒ Tubing; 0.32 cm ID, 0.64 cm OD)
installed adjacent to three of the five headspace chambers in each
polygon at depths of 5, 10, 15, 20, and 30 cm (or to the depth of icecemented soil). The exposed tube above the soil surface was sealed
with an injectable septum (Baxter Interlink Injection Site Ref
#2N3399K); the buried base of the subsurface sampling tube was
covered with fine polyester mesh to prevent soil particle ingress
during CO2 sampling. Gas samples from each of the subsurface
sampling tubes were taken during the surface soil CO2 fluxsampling period. Residual air within each sampling tube was
extracted and discarded before a 20 ml sample was taken.
Soil CO2 sampling began >96 h after installation of equipment.
Gas samples were taken during the warmest and coolest periods of
the day, based on surface air temperature measurements at each
site, on two days in late December 2008. Site B was only sampled
during one “cold” period due to an error in determining the timing
of the coldest part of the day. All CO2 samples were taken using gastight syringes (BD 30 ml syringe; Ref 309650) fitted with 3-way
stopcocks (Baxter 2C6201) and 26G ½” needles (BD PrecisionGlide). Samples were injected into evacuated (0 atm) 12 ml
vials (ExetainersÒ, Labco Inc., UK) fitted with additional polytetrafluoroethylene (PTFE) septa (Supelco Analytical, USA) to maintain
sample integrity (Rochette and Bertrand, 2003; Knohl et al., 2004),
and stored in insulated containers prior to transport back to New
Zealand. All samples were analysed for CO2 concentration and
d13 CCO2 using a trace gas module (TGII, Sercon Ltd., Cheshire, UK)
connected to a continuous flow isotope ratio mass spectrometer
(Sercon Ltd., Cheshire, UK). The d13C values are expressed in parts
per thousand (&) enrichment or depletion of 13C, relative to Vienna
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Pee Dee Belemnite (VPDB). Soil profile CO2 samples were analysed
at Lincoln University, New Zealand (within 4 months of being
collected), and surface CO2 flux samples at the University of California Davis Stable Isotope Facility, California, USA (within 8
months of being collected). Standard reference gas samples were
also collected in the field following surface CO2 flux and subsurface
CO2 sampling, and were analysed with the samples. The standards
showed no sign of storage effects. The analytical precision of CO2
concentration and d13C values was 5 ppm and 0.1&, respectively.
2.4. Soil sampling, physical and chemical analyses
In polygons adjacent to CO2 sampling sites, soil pits were dug by
hand to ice cement, and deepened to a minimum of 55 cm depth
using a concrete breaker to penetrate ice-cemented material. Soils
were described and classified using USDA nomenclature before
being volumetrically sampled (500 cm3) at 5 cm contiguous
intervals to 50 cm depth. All samples were returned to New Zealand
where they were air-dried, and weighed to determine bulk density
(rb) before being sieved (<2 mm). All chemical analyses were
performed in duplicate on the <2 mm fraction.
Soil pH and electrical conductivity were determined on a 1:5
soil:deionised water suspension (United States Salinity Laboratory
Staff, 1954). Samples were subsequently filtered through 0.45 mm
syringe filters (Advantec MFS-25) to remove suspended sediment
prior to analysing for cations and anions. Cations were determined
by inductively coupled plasma optical emission spectroscopy (720ES, Varian, Melbourne, Australia) and anions by ion exchange
chromatography (DX-120, Dionex, Sunnyvale, California, USA). Total
C and nitrogen (N) contents of solid samples were determined on
a finely ground (<50 mm) subsample of the <2 mm fraction. Samples
were combusted at 1000 C using a Europa ANCA-GSL CN analyser
connected to a continuous flow isotope ratio mass spectrometer
which also determined d13C and d15N values (Sercon Ltd., Cheshire,
UK). The d13C and d15N values are expressed relative to VPDB and
atmospheric N2, respectively. Two separate runs were performed in
order to obtain total soil d13C and organic d13C values, and to enable
calculation of soil inorganic C content. Samples analysed for total
soil C, N, d13C and d15N were not subjected to any treatment.
Samples analysed for organic d13C and total organic C content were
fumigated with HCl to remove all inorganic C prior to analysis.
Subsamples of the finely ground fraction were put into small glass
vials, moistened with deionised water, and placed in a vacuum
desiccator with a beaker containing 100 ml of concentrated (12 M)
HCl for 12 h (Harris et al., 2001). Samples were then dried at 60 C,
cooled in a desiccator and capped tightly prior to being weighed into
tin capsules for analysis. Soil inorganic C contents were calculated
by subtracting the total C content of the HCl fumigated samples from
the total C values obtained from the untreated samples.
2.5. Statistical analyses
Principal component analysis (PCA) was used to identify the
main components influencing the variance in soil chemical properties with depth and between sites. PCA was completed using the
function ‘princomp’ in R 2.11.1 (The R Foundation). Linear regression was used in Microsoft ExcelÒ 2007 to determine surface CO2
13
fluxes and the rate of change in d CCO2 in surface flux samples.
3. Results
101
although soil temperatures during the sampling periods were very
similar at both sites (Figs. 1A and 2A). The strongest vertical soil
temperature gradients occurred during “warm” sampling periods,
with soil temperatures reaching 13e14 C at 5 cm depth, and
1.3e2.6 C at 25 cm depth. During “cold” sampling periods, soil
temperature gradients were lower, with temperatures ranging
from w0.5 C at 5 cm depth to w1.6e3.6 C at 25 cm depth. The
progressively greater lag in soil temperature response to surface
heating with increasing depth in the profile meant that soil
temperatures at 25 cm depth during the “cold” sampling period
were slightly warmer than those measured during the “warm”
sampling period. The rate of change in soil temperature during
warm and cold sampling periods was always greatest at 5 cm
depth, and decreased with depth in the soil profile.
Soil moisture content remained relatively constant throughout
the sampling period at Site A (Fig. 1B), with no moisture detected at
5 cm depth. Ponded water was observed on the soil surface in lowlying areas within 300 m of the sampling area, suggesting subsurface melt from ice-cemented soils on adjacent north-facing hill
slopes was contributing to the relatively high moisture content
(w5%) observed at 15 cm depth at this site. Soil moisture content at
25 cm depth was around 6%. At Site B, soil moisture content at 5 cm
depth was higher during warm sampling periods, whereas at 15 cm
depth it remained relatively constant (w1%) (Fig. 2B). Soil moisture
content at 25 cm depth increased from 4 to 8% over the sampling
period due to thawing ice cement at the base of the profile.
3.2. Soil physical and chemical properties
Soil bulk density ranged from 1.5 to 2.3 g cm3. These values
correspond to soil porosities ranging from 18 to 45%, assuming
a particle density of 2.8 g cm3 (Barrett and Froggatt, 1978) which is
appropriate for the dominantly mafic igneous rocks of the study area.
Principal component analysis of soil chemical data revealed
three principal components which explained 80% of the variance in
the original soil chemical data. Principal component (PC) 1 represents total C, inorganic C, calcium (Ca2þ), magnesium (Mg2þ),
2
nitrate (NO
3 ) and sulphate (SO4 ) (Supplementary Table 1). PC 1
can be considered a C component as it is dominated by total C and
inorganic C. PC 2 represents pH, EC and sodium (Naþ), and is
effectively a measure of saltiness. PC 3 represents total N, organic C,
potassium (Kþ), chloride (Cl) and sulphate, although component
correlations suggest that total N and organic C are the variables best
described by PC 3. Depth profiles, contrasts and similarities
between the soils at each site are discussed below in terms of the
major variables contributing to the three principal components.
3.2.1. Inorganic C (PC 1)
Inorganic C was higher at Site A than at Site B (Fig. 3A). At Site A,
inorganic C decreased from 2 mg g1 to 1 mg g1 in the upper 15 cm
of the profile, and increased with depth below 15 cm. The
maximum inorganic C content (4 mg g1) occurred in the
40e45 cm depth increment, coinciding with the approximate
maximum thaw depth of the active layer. Despite inorganic C
content in the upper 15 cm being the lowest in the profile at Site A,
it was more than double the organic C content. Below 15 cm depth,
inorganic C was always more than four times greater than organic C
at the same depth. At Site B, inorganic C decreased from 2 mg g1 in
the top 5 cm to 0.3 mg g1 between 10 and 25 cm depth. Below
25 cm depth, inorganic C increased, and was always more than five
times greater than organic C at the same depth.
3.1. Environmental variables
At the time of sampling (late December 2008), soils at Sites A
and B were ice-cemented below 35 and 26 cm, respectively,
3.2.2. Organic C (PC 3)
Total organic C decreased with depth down the profile at Site A
(Fig. 3B), except for a secondary peak of 0.6 mg g1 at 40e45 cm
102
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Fig. 1. Soil temperature (A) and volumetric moisture content (B) during gas sampling periods, Site A Taylor Valley. Note: data at 5 cm depth were only recorded during one sampling
period due to a faulty probe.
depth, which was only slightly lower than values measured
between 0 and 10 cm depth. The depth of the secondary peak
corresponds to the estimated maximum thaw depth of the active
layer. At Site B, total organic C values were very low
(0.16e0.32 mg g1) and decreased slightly with depth.
From 5 to 15 cm depth at Site A, and in the upper 10 cm of the
profile at Site B, d13C values were highly depleted relative to VPDB,
ranging from 25.7 to 30.7& (Fig. 3C). The upper 5 cm at Site A
had a comparatively enriched d13C value of 21.6&; below 15 cm
depth, values ranged from 15.8 to 19.6&. At Site B, d13C values
beneath 10 cm depth were more depleted relative to those at Site A,
ranging from 18.6 to 20.6&.
3.2.3. Soil pH and electrical conductivity (PC 2)
Soil pH ranged from w10 in the upper part of the profile at both
sites to between 8.8 and 9.4 at 45e50 cm depth (Fig. 4A). Electrical
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
103
Fig. 2. Soil temperature (A) and volumetric moisture content (B) during gas sampling periods, Site B Taylor Valley. Note: this site was only sampled during one “cold” period due to
an error in determining the timing of the coldest part of the day.
conductivity (EC), a proxy for soil salt concentration, was higher in
the upper 10 cm of the soil at Site B than at Site A (Fig. 4B). At both
sites, EC values were relatively low and constant between 10 and
40 cm depth. Between 40 and 50 cm depth, EC values at both sites
increased to a level similar to that in the upper 5 cm at Site A
(Fig. 4B).
3.3. CO2 dynamics
3.3.1. Subsurface CO2 profiles
Subsurface CO2 concentrations were significantly different
between warm and cold periods at both sites (Fig. 5A). During
warm periods, soil CO2 concentrations were greater than
104
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Fig. 3. Depth profiles of A) soil inorganic C, B) soil organic C, and C) C isotopic composition of soil organic C at Sites A and B, Taylor Valley (error bars are 1 standard deviation; n ¼ 2).
atmospheric CO2 concentration (383.3 mL L1 at the South Pole in
December 2008; Keeling et al., 2001) in the upper 10 cm of the
profile at Site A, reaching a maximum of 400 mL L1 at 10 cm depth.
Samples taken during cold sampling periods showed soil CO2
concentrations were substantially below that of atmospheric CO2,
the lowest being 315 mL L1 at 15 cm depth at Site A. At both sites,
shifts in soil CO2 concentration between warm and cold periods
were mirrored about a trend-line that decreases from 380 mL L1 at
the surface to 330e340 mL L1 at 30 cm depth, demonstrating that
cycles of CO2 uptake (cold periods) and release (warm periods)
occur. The amount of soil CO2 taken up and released was greater at
Site A than at Site B.
13
Soil profile d CCO2 was relatively enriched during warm periods
when compared to cold periods at both sites (Fig. 5B). At Site A,
d13 CCO2 became more enriched with increasing depth in the profile
during both warm and cold periods, although at 15 cm depth there
were slight inflexions in the trends. Values ranged from 7.0
to 9.0&, with a maximum difference between warm and cold
13
periods of 1&. At Site B, d CCO2 remained relatively constant with
depth during both warm and cold periods, ranging from 8.2
to 9.2&, with differences of 0.9& between warm and cold
13
periods. Soil profile d CCO2 was similar to that of atmospheric CO2
(8.22& at the South Pole in December 2008; Keeling et al., 2001)
13
during warm periods at Site B. At both sites, d CCO2 was highly
enriched relative to soil organic matter (Fig. 3C).
3.3.2. Surface CO2 fluxes
flux
rates
ranged
from
0.050
to
Surface
CO2
0.048 mmol m2 s1 (Table 1). Periods of efflux (positive flux) and
influx (negative flux) corresponded to warm and cold sampling
periods, respectively. During warm periods (efflux), headspace CO2
became progressively enriched (by up to 1.8&), whereas during
cold periods (influx), headspace CO2 became depleted (by up
13
to 2.0&). The d CCO2 values ranged from 6.5 to 10.6& during
warm periods, and 7.4 to 11.2& during cold periods.
4. Discussion
4.1. Soil CO2 uptake and release
The low soil CO2 flux rates measured in this study are in close
agreement with in situ surface flux rates in Taylor Valley reported
by Burkins et al. (2001) and Parsons et al. (2004), but an order of
magnitude lower than those reported by Ball et al. (2009), despite
Fig. 4. Depth profiles of A) soil pH, and B) electrical conductivity at Sites A and B, Taylor Valley (error bars are 1 standard deviation; n ¼ 2).
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
105
13
Fig. 5. Soil profile CO2 concentration (A) and d CCO2 (B) at Sites A and B, Taylor Valley. Data shown represent averages for warm and cold sampling periods (n ¼ 6 for 0 cm samples
and n ¼ 12 at all other depths, except for the cold sampling period at Site B, where n ¼ 3 for 0 cm samples and n ¼ 6 at all other depths). Error bars represent the SEM.
all studies having used soils with similar pH, EC, organic and
inorganic C values. Higher flux rates such as those reported by
Gregorich et al. (2006) were from soils with organic C contents of
up to 35 mg g1. Such high organic C contents are unusual in
Antarctic dry valley soils; this study focuses on CO2 dynamics in
typical dry valley soils characterised by high pH (w10) and low
organic C contents (<1 mg g1).
It is well established that soil respiration increases with rising
soil temperature (e.g. Lloyd and Taylor, 1994; Fang and Moncrieff,
2001). If CO2 release during warm periods at our study sites were
13
due to biological respiration, d CCO2 values should reflect the d13C
of the source organic material (Dörr and Münnich, 1980). Soil
d13 CCO2 values (Fig. 5B) showed no relationship with soil organic
matter d13C values (Fig. 3C). Furthermore, if CO2 uptake during
cold periods were a result of algal photosynthesis, we would
13
expect soil d CCO2 to be enriched due to preferential uptake of
12
CO2 by photosynthetic organisms, thus concentrating the
heavier 13CO2 in the soil atmosphere (Park and Epstein, 1960). No
biological mechanism can be invoked to account for our results.
13
Instead, d CCO2 values associated with the uptake (depleted
13
d CCO2 ) and release (enriched d13 CCO2 ) of soil CO2 indicate that
CO2 dynamics in Antarctic soils are controlled by other
mechanisms.
In studies of surface CO2 fluxes in Taylor Valley, Ball et al. (2009)
found that soil temperature fluctuations (at 5 cm depth) accounted
for almost half of the variation observed in diel CO2 fluxes. As
similar diel cycles of CO2 flux were also observed in heat-treated
(120 C) soils, presumed to be free of biological activity, Ball et al.
106
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Table 1
13
Surface soil CO2 fluxes and rate of change in d CCO2 during surface CO2 flux
sampling at Sites A and B, Taylor Valley.
Site
Chamber
Sampling timea
CO2 flux (mmol m2 s1)
Dd13 CCO2 b
Dt
A
a
W1
W2
C1
C2
0.010
0.000
0.041**
0.003
0.010
0.013*
0.006
0.008
A
b
W1
W2
C1
C2
0.034
0.026
0.024
0.044
0.022**
0.014
0.001
0.002
A
c
W1
W2
C1
C2
0.048
0.019
0.039**
0.038**
0.000
0.003*
0.008
0.004
A
d
W1
W2
C1
C2
0.043**
0.005
0.004
0.016
0.024**
0.016
0.022**
0.016*
A
e
W1
W2
C1
C2
0.034*
0.024*
0.001
0.039**
0.003
0.007
0.051
0.026**
A
f
W1
W2
C1
C2
0.036
0.016**
0.023**
0.024*
0.008
0.029
0.011**
0.021*
B
a
W1
W2
C1
0.025
0.034**
0.037
0.015
0.010
0.018
B
b
W1
W2
C1
0.021
0.018**
0.028*
0.021
0.011
0.025
B
c
W1
W2
C1
0.045
no data
0.050**
0.002
no data
0.023
B
d
W1
W2
C1
0.040*
0.010
0.013
0.012
0.020*
0.013
B
e
W1
W2
C1
0.018
0.018**
0.029*
0.011
0.009
0.008*
B
f
W1
W2
C1
0.024*
0.013
0.001
0.004
0.011
0.004
* represents P < 0.1; ** represents P < 0.05.
a
W and C denote warm and cold sampling periods, respectively.
b
Units are per mil min1.
(2009) suggested that dissolution of CO2 in soil water, consistent
with Henry’s Law, was a likely explanation for the diel variations
observed. We test this hypothesis by: 1) comparing soil profile CO2
concentrations measured at warm and cold periods with values
simulated by a model that incorporates Henry’s Law dissolution of
13
CO2, and 2) comparing the measured changes in d CCO2 with those
expected from fractionation due to dissolution and exsolution of
CO2, biological respiration, and photosynthetic CO2 fixation. We
also use our data to quantify the relative contributions of biotic and
abiotic processes to soil CO2 uptake and release.
4.1.1. A model of abiotic CO2 dynamics
At steady state, the closed-system partitioning of CO2 between
gaseous and dissolved CO2 is determined by Henry’s Law and
solution pH. The temperature dependent Henry coefficient (KH)
(Stumm and Morgan, 1996) expresses the relationship between
pCO2 of the atmosphere and dissolved gaseous CO2
([CO2(aq)] ¼ [H2CO3]), as pCO2 ¼ KH[CO2(aq)]. Total dissolved CO2
2
(CT), which includes the species H2CO3, HCO
3 and CO3 , is related to
[CO2(aq)] and pH by:
h
i
Ka
Ka Ka
CT ¼ CO2ðaqÞ 1 þ þ1 þ 1 22
H
Hþ
!
(2)
where Ka1 and Ka2 are the equilibrium constants for [Hþ] [HCO
3 ]/
[H2CO3] and [Hþ] [CO2
3 ]/[HCO3 ], respectively (Stumm and Morgan,
1996). The temperature dependence of KH is such that as temperature increases, [CO2(aq)] decreases. Thus, as a result of changes in
KH, decreasing temperature causes CO2 absorption, and increasing
temperature causes CO2 desorption. In a soil, the capacity of soil
solution to absorb and desorb CO2 depends on soil solution volume,
soil pH, and the magnitude of soil temperature variation. From Eq.
(2), it is evident that increasing soil solution pH (decreasing [Hþ])
causes an increase in CT. A high soil solution pH interacts with the
volume of soil solution, amplifying the effect of CO2 adsorption or
desorption with changing temperature by providing for a larger
source or sink of dissolved CO2 involved in the exchanges.
Using the principles described above, and closed-system, steady
state assumptions, we model abiotic CO2 dynamics in the soils at our
study sites. The low flux rates measured (Table 1) suggest the closedsystem assumption is not unreasonable, but we explore this further
below. The model predicts the ratio of soil CO2 concentrations
([CO2]warm:[CO2]cold) when temperature (T) drops from warm
(Twarm) to cold (Tcold) in terms of the soil moisture content and pH of
the system.
The ratio of subsurface soil CO2 concentrations attributable to
abiotic processes as a result of a change in temperature from Twarm
to Tcold is given by:
½CO2 warm
½CO2 cold
q
ð4 qÞ
þ
KHcold
RTcold
T
cold
¼ q
Twarm
ð4 qÞ
þ
RTwarm KHwarm
(3)
where 4 is soil porosity (L voids L soil1), q is soil volumetric
moisture content (L water L soil1), R is the ideal gas constant
(8.314 J K1 mol1), T is soil temperature (K), KH is Henry’s coefficient (L atm mol1) and [CO2] is soil CO2 concentration (mL L1).
A full derivation of the model is given in Appendix A.
The dependence of [CO2]warm:[CO2]cold on soil pH and moisture
content for temperature changes (between warm and cold sampling
periods) typical of those measured at 5 cm (Fig. 6A) and 15 cm
(Fig. 6B) depth in the study area is shown in Fig. 6. At low pH (6), CT
is low (Eq. (2)), and consequently a large soil solution volume (high
moisture content) is required to provide the pool of dissolved inorganic C necessary for temperature-driven absorption/desorption
processes to shift [CO2]warm:[CO2]cold significantly. Between pH 7
and 9, CT rises rapidly and the volume of soil solution required to
produce a given [CO2]warm:[CO2]cold declines rapidly. For example, at
pH 6, soil moisture content must be 0.32 (at or above field capacity in
many soils) for [CO2]warm:[CO2]cold to be 1.4 (Twarm ¼ 286 K and
Tcold ¼ 274 K; Fig. 6A). By contrast, for the same change in T at pH 9,
the same ratio can be accommodated by a soil with a moisture
content < 0.01 (<1%). The magnitudes of [CO2]warm:[CO2]cold simulated for 15 cm depth (Fig. 6B) are much lower because of the smaller
temperature fluctuation at this depth, but the relative sensitivity to
soil pH and moisture content remains the same.
The model suggests that changes in soil profile CO2 concentrations in the study area should be greatest at shallow soil depths,
where temperature changes are greatest, pH is high (w10), and
moisture content is adequate. Conversely, greater depths should
have smaller changes in soil CO2 concentrations due to smaller
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
107
Fig. 6. Ratios of simulated [CO2]warm:[CO2]cold as influenced by soil moisture content and soil solution pH. Ratios are calculated using Eq. (3). In (A) Twarm ¼ 286 K and Tcold ¼ 274 K
which are similar to soil temperatures measured at 5 cm depth, and in (B) Twarm ¼ 278.3 K and Tcold ¼ 277.3 K, similar to soil temperatures measured at warm and cold periods,
respectively, at 15 cm depth in the study area. When the ratio of [CO2]warm:[CO2]cold is 1, there is zero potential for CO2 to be absorbed. This occurs when soil moisture is so low that
no CO2 can be dissolved. Higher ratios represent greater changes in soil CO2 concentrations between warm and cold periods, and thus greater uptake and release of CO2 by soils.
changes in soil temperature. However, our measured CO2 concentration depth profiles (Fig. 5A) show relatively low variation in CO2
concentrations at shallow depths, with maximum differences at
10e15 cm. As predicted by the model, little variation occurs at the
base of the profile.
Measured and simulated ratios of [CO2]warm:[CO2]cold were
compared for 5, 15 and 25 cm depths using measured values of soil
pH, temperature and volumetric moisture content (Fig. 7). The
25 cm depth increment behaves as the model predicts. Ratios of
[CO2]warm:[CO2]cold are very close to 1 due to minimal changes in
soil temperature between warm and cold periods. At this depth, the
soil closely approximates a closed system because ice-cemented
soil beneath provides a gas-impermeable lower barrier, and CO2
concentration gradients are low (Fig. 5A). At 15 cm depth, measured
ratios are 15% and 9% greater than simulated ratios (their theoretical abiotic maximum) at Sites A and B, respectively, suggesting
a possible biological respiration component. At 5 cm depth,
measured [CO2]warm:[CO2]cold ratios are considerably lower than
simulated ratios. As the 5 cm depth increment is most likely to
suffer from violation of the closed-system assumption of the model,
108
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Fig. 7. Comparison between measured and simulated [CO2]warm:[CO2]cold ratios between warm and cold sampling periods for Sites A and B, Taylor Valley.
we attribute this discrepancy to exchange between the soil and
surface atmospheres. Leakage of CO2 from the soil during warm
periods when CO2 is released via exsolution, and influx of atmospheric CO2 during cold periods when soil CO2 concentrations are
depleted (CO2 dissolution) would act to damp variation in
measured CO2 concentrations. These inferred fluxes are supported
by the soil CO2 concentration gradients at this depth (Fig. 5A).
4.1.2. Discriminating between biotic and abiotic processes using
stable C isotopes
Biotic production and consumption of CO2 and the abiotic dissolution/exsolution processes outlined above produce distinctive soil
d13 CCO2 values. Biological respiration of highly depleted organic
C would produce respired CO2 of a very similar (highly depleted) C
isotopic composition (Dörr and Münnich, 1980), as respiration does
not (or only to a minor extent) discriminate against 13C. Based on
measured soil organic d13C values at both sites (Fig. 3C), we would
13
expect respired d CCO2 values of w 30 to 21& in the upper 15 cm
of the profiles, and w 20 to 16& below 15 cm depth. Thus, if
biological respiration were responsible for the increase in soil CO2
13
concentration during warm periods, d CCO2 values would become
relatively depleted. Conversely, discrimination against 13CO2 in C3
13
photosynthetic pathways would produce a d CCO2 -enriched atmosphere above actively photosynthesising organisms. Therefore, if soil
CO2 uptake during cold periods were due to photosynthetic CO2
Table 2
Carbon isotope fractionation in the CO2eHCO3eCO3eCaCO3 system under equilibrium conditions. 133 y/x represents the fractionation of compound y relative to
compound x. Soil temperatures (T) shown are similar to those measured during the
sampling period. g ¼ gaseous CO2, a ¼ dissolved CO2, b ¼ dissolved HCO
3,
c ¼ dissolved CO2
3 ions, s ¼ solid calcium carbonate. TK ¼ T ( C) þ 273.15 K.
T ( C)
13
a
3 g/a
1
5
1.17
1.15
(&)
13
3 g/b
b
(&)
10.70
10.20
13
c
3 g/c
(&)
10.06
9.61
a 13
3 g/a ¼ 0.19& þ 373/TK (Vogel et al., 1970).
b 13
3 g/b ¼ 23.89& 9483/TK (Mook et al., 1974).
c 13
13
13
3 g/c ¼
3 g/b þ
3 b/c (¼ 2.52& þ 867/TK; Thode
d 13
3 g/s ¼ 9.15& 5380/TK (Mook, 2000).
et al., 1965).
13
3 g/s
d
(&)
10.47
10.19
13
consumption, d CCO2 values would become relatively enriched. Our
13
results are contrary to these patterns: soil profile d CCO2 values were
relatively enriched during warm periods (CO2 release) and depleted
during cold periods (CO2 uptake; Fig. 5B). Surface CO2 flux data also
13
showed progressive enrichment and depletion of d CCO2 during
warm and cold sampling periods, respectively (Table 1). The observed
patterns are consistent with isotopic fractionation during CO2 disso2
lution and its speciation to HCO
3 (aq) and CO3 (aq) (Table 2). At pH
values >9, dissolved CO2 species are dominated by the HCO
3 and
CO2
3 forms (Stumm and Morgan, 1996). Equilibrium fractionation of
2
CO2 gas to HCO
3 and CO3 results in a w 10& depletion of CO2 gas
2
relative to HCO3 or CO3 (Thode et al.,1965; Mook et al.,1974; Table 2).
2
Thus, dissolution of CO2 to form HCO
3 and CO3 (at the high pH values
13
measured) would result in depletion of d CCO2 in the soil atmo13
sphere, whereas exsolution would lead to enrichment of d CCO2 in
the soil atmosphere. The magnitudes of the isotopic shifts we observe
(w 1&) are less than w 10& because only a proportion of the CO2
in the soil atmosphere exchanges with soil solution (w10%).
13
Although the d CCO2 data confirm abiotic processes dominate
CO2 dynamics in soils of Taylor Valley, as postulated by Ball et al.
(2009), our comparison of measured and simulated [CO2]warm:[CO2]cold ratios suggests there is a component of CO2 flux around
15 cm depth that cannot be accounted for by abiotic processes alone.
We use soil organic d13C values in conjunction with soil profile
d13 CCO2 values to further constrain the potential biotic contribution
to soil CO2 fluxes.
4.1.3. Quantifying the biotic contribution to soil CO2 fluxes
The bulge in soil CO2 concentration from 5 to 15 cm depth at Site
A is likely to have a biological respiration component, as organic C
content at this depth is relatively high (w0.6 mg C g soil1; Fig. 3B)
and the soil is relatively warm (Fig. 1A). Assuming respired soil CO2
13
at this depth would have an average d CCO2 value of 28.2&
(Fig. 3C), we can use a mixing model (Fry, 2006) to calculate the
biotic contribution (fbiotic) to soil CO2 fluxes:
fbiotic ¼
d13 CCO2 produced d13 CCO2 abiotic
d13 Cbiotic d13 CCO2 abiotic
(4)
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Table 3
Proportion of the change in CO2 concentration and isotopic composition attributable
to biotic processes (fbiotic) at Site 1, 5e15 cm depth. This depth has the greatest shift
in CO2 concentration between cold and warm periods (Fig. 5A) and is the most likely
place in which biological respiration may occur. The effects of changes in assumed
d13 CCO2 abiotic values on fbiotic are shown.
fbiotic
d13 CCO2 abiotic (&)
13
Source of d CCO2 abiotic value
0.15
8.22
Atmospheric d CCO2 at the South Pole in
December 2008 (Keeling et al., 2001).
0.05
4.10
Average d CCO2 emitted from the whole
profile at Site 3, assuming no biological
contribution.
0.14
1.42
Assumes Site 3 has a 10% biological
contribution, as suggested by the
comparison between measured and
simulated [CO2]warm:[CO2]cold ratios
(Fig. 7). The 10% biological contribution
is conservatively applied to the whole profile.
0.24
2.16
13
13
13
Assumes no fractionation of CO2 as it is
exsolved from a 50:50 mix of
2
HCO
3 :CO3 in isotopic equilibrium
with atmospheric CO2 at 1 C (Table 2).
13
where d CCO2 produced is the d CCO2 (&) produced from cold to
13
13
warm periods (Appendix B), d CCO2 abiotic is the d CCO2 (&) of CO2
13
produced due to abiotic processes, and d Cbiotic is the d13C (&) of
the organic source material (Fig. 3C).
This calculation requires that we know the isotopic composition
13
of abiotically produced CO2 (d CCO2 abiotic ). We calculate fbiotic using
13
a range of d CCO2 abiotic values (Table 3) appropriate to different
assumptions about isotopic fractionation during dissolution and
exsolution of CO2. The proportion of biotic CO2 production increases
13
with increasing values of d CCO2 abiotic . The most depleted value of
13
d CCO2 abiotic (8.22&) assumes exsolved CO2 has the same d13 CCO2
as atmospheric CO2 at the time of sampling. This is clearly unrealistic, because firstly, it produces a negative fbiotic value, and secondly,
13
exsolved CO2 this depleted would not allow a shift of soil d CCO2
13
to > 8.22& as is measured at Site A (Fig. 5B). A d CCO2 abiotic value
of 4.1& suggests a biotic contribution of 5%. This value of
d13 CCO2 abiotic is determined using a mass balance calculation of the
13
C isotopic composition necessary to produce the shift in soil d CCO2
from cold to warm sampling periods at Site B (see Appendix B for
details of the calculation). Data from Site B are used because the low
organic C status of the soil is likely to limit respiration to the extent
that CO2 uptake and release throughout the whole profile is dominated by abiotic processes. However, our model of abiotic soil CO2
uptake and release suggests there may be as much as a 9% biological
component to soil CO2 production at 15 cm depth at Site B. Therefore,
conservatively assuming a 10% biological contribution from respiration of soil organic C to soil CO2 production over the whole profile
13
at Site B, the value of d CCO2 abiotic is 1.42& (Appendix B), which
results in a biological respiration component of 14%. The highest
13
value of d CCO2 abiotic assumes no fractionation of CO2 released by
2
exsolution from a 50:50 mix of HCO
3 and CO3 in isotopic equilibrium with atmospheric CO2. This is an unlikely scenario, but it
13
establishes the most enriched possible value of d CCO2 abiotic . The
resultant biotic contribution to soil CO2 fluxes is 24%.
4.1.4. Implications for soil respiration measurements and C turnover
times
Long-term records of soil respiration in the McMurdo Dry
Valleys have been suggested as being one of the most sensitive
indicators of ecosystem response to climate variability (Barrett
et al., 2006c). However, it is essential that both biotic and abiotic
contributions to soil CO2 fluxes are identified and measured
109
accurately, as both components are highly spatially and temporally
variable. The techniques we have utilised represent a first attempt.
Our data demonstrate that previously measured in situ soil CO2
fluxes attributed to biological respiration may be significant overestimates (e.g. Burkins et al., 2001; Barrett et al., 2006b; Elberling
et al., 2006; Gregorich et al., 2006), and C turnover times calculated on the basis of such fluxes may be significant underestimates
(e.g. Burkins et al., 2001; Elberling et al., 2006). Based on their
calculation of a mean organic C turnover time of 23 years, Burkins
et al. (2001) inferred that C cycling in Taylor Valley is non-steady
state, and suggested that at least two organic C pools contribute
to C cycling: a labile organic C pool replenished by in situ photosynthesis, and a more recalcitrant organic C pool derived from
ancient glacial and lake sediments. Such a short C turnover time
implies relatively low ecosystem resilience to fluctuations in C
supply. Our results, which show that biological respiration accounts
for a (conservative) maximum of 25% of measured soil CO2 fluxes in
the Dry Valleys environment, mean that present estimates of C
turnover times could be at least four times greater, in the order of
100e500 years. The increased turnover times resulting from lower
rates of C utilisation imply that Dry Valley ecosystems are likely to
be more resilient to environmental change than what present
estimates of C turnover time suggest.
5. Conclusions
Soil CO2 dynamics at the two sites studied in Taylor Valley are
dominated by abiotic processes, with small changes in soil
temperature capable of driving significant changes in soil CO2
concentration. The isotopic data imply that biological respiration
makes only a minor contribution to soil CO2 fluxes in the Dry
Valleys environment.
Our results have significant implications for understanding C
cycling in polar desert environments. The abiotic influence on soil
CO2 dynamics must be accounted for and monitored, as CO2
effluxes cannot be assumed to be biologically driven.
Acknowledgements
We thank Prof. James Bockheim for his assistance and enthusiasm in the field, Neil Smith for his expert technical help with data
logger and field equipment set-up, and Roger Cresswell for his
invaluable assistance with sample analyses. We gratefully
acknowledge the excellent logistical support provided by
Antarctica New Zealand. Operational support was provided by the
Lincoln University Research Fund. F.L.S thanks Helicopters New
Zealand, the William Machin Trust, Freemasons New Zealand and
the Kate Sheppard Memorial Trust for financial support. We also
thank two anonymous reviewers for their constructive comments
on the manuscript.
Appendices
Appendix A. Model of abiotic CO2 uptake and release
Model calculations were based on the following equations.
Assuming a soil at steady state, at temperature T (K) for an increment of depth D (m), the partial pressure of CO2, according to
Henry’s Law, is
h
i
pCO2 ¼ KHT CO2ðaqÞ ;
(A.1)
where KH is the Henry coefficient (L atm mol1) and [CO2(aq)] is
dissolved gaseous CO2 (H2CO3).
110
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Total dissolved CO2 (CT) is determined by the equilibrium
2
þ
constants for [Hþ] [HCO
3 ]/[H2CO3] and [H ] [CO3 ]/[HCO3 ], Ka1 and
Ka2 , respectively.
ZD
i
h
CT ¼ CO2ðaqÞ
0
Ka
Ka Ka
1 þ þ1 þ 1 22
H
Hþ
!
¼
pCO2
P½CO2 ¼
;
0
0
KH
KH
T
T
(A.2)
0 ¼ K ð1 þ ðK =½Hþ Þ þ ðK K =½Hþ 2 ÞÞ, [CO ] is the
where KH
a1
a1 a2
2
HT
T
concentration of soil CO2 (mL L1) and P is the pressure of the soil
atmosphere (atm).
Per unit area of soil surface, the number of moles of dissolved
CO2 (mmol) is given by
nCO2dissolved ¼
P½CO2 Dq;
0
KH
T
(A.3)
where q is the soil volumetric moisture content (L water L soil1).
The number of moles of CO2 in the soil atmosphere (mmol)
nCO2soil atmos ¼
½CO2 PVsoil atmos
P½CO2 Dð4 qÞ
¼
;
RT
RT
(A.4)
where Vsoil atmos is the volume of the soil atmosphere (L), and 4 is
soil porosity (L voids L soil1).
At temperature T1, total moles of CO2 in volume D
d13 CCO2 cold ½CO2 cold dz þ d13 CCO2 produced ½CO2 produced
ZD
¼
(B.1)
13
d CCO2 warm ½CO2 warm dz;
0
where z is soil depth (m), D is the depth to ice cement (m), [CO2] is
13
soil CO2 concentration (mL L1) and d CCO2 produced is the
13
concentration-weighted mean d CCO2 of the produced CO2 (&).
The term ½CO2 produced was calculated as
ZD
ZD
½CO2 warm dz 0
½CO2 cold dz:
(B.2)
0
Assuming there is no biological contribution to the CO2
produced at Site B,
d13 CCO2 abiotic ¼ d13 CCO2 produced :
(B.3)
If, however, there is a 10% biological contribution to the CO2
13
produced at Site B, and assuming that d CCO2 reflects the d13C of
P1 ½CO2 T1 Dð4 qÞ P1 ½CO2 T1
¼
þ
Dq
0
RT1
KH
T1
!
q
ð4 qÞ
:
¼ P1 ½CO2 T1 D
þ 0
RT1
KHT
(A.5)
13
the organic source material (d13Cbiotic), then d CCO2 abiotic (the
13
mean d CCO2 (&) of CO2 produced due to abiotic processes) can be
derived as
1
Under the closed-system assumption, conservation of mass
dictates that between T1 and T2,
P2 ½CO2 T2 D
q
ð4 qÞ
þ 0
RT2
KHT
!
¼ P1 ½CO2 T1 D
2
q
ð4 qÞ
þ 0
RT1
KHT
!
;
1
(A.6)
13
d CCO2 abiotic ¼
d13 CCO2 produced 0:1 d13 Cbiotic
0:9
:
(B.4)
13
d CCO2 abiotic
Values
used
to
calculate
are
13
[CO2]produced ¼ 29 mL L1, d CCO2 produced ¼ 4.1& and
d13Cbiotic ¼ 28.2&. The integrals were approximated from the
point data shown in Fig. 5.
and therefore
P1 ½CO2 T1
¼
P2 ½CO2 T2
Appendix C. Supplementary material
!
q
ð4 qÞ
þ 0
KHT
RT2
2
q
ð4 qÞ
þ 0
RT1
KHT
1
!:
(A.7)
References
Given the exchanges of CO2 between atmosphere and solution
phases are small in relation to the total mass of the soil atmosphere,
from the ideal gas law, the ratio of the pressures of the soil atmosphere between T1 and T2 is equal to the ratio of the temperatures,
and therefore
½CO2 T1
¼
½CO2 T2
q
ð4 qÞ
þ 0
KHT
RT2
2
q
ð4 qÞ
þ 0
RT1
KHT
Supplementary material associated with this article can be
found, in the online version, at doi:10.1016/j.soilbio.2012.04.027.
!
!
T2
:
T1
(A.8)
1
Appendix B. Calculating the isotopic composition of abiotically
13
produced CO2 (d C CO2 abiotic )
13
Using a mass balance approach, the average d CCO2 of the CO2
produced between cold and warm sampling periods must satisfy
the following
Adams, B.J., Bardgett, R.D., Ayres, E., Wall, D.H., Aislabie, J., Bamforth, S., Bargagli, R.,
Cary, C., Cavacini, P., Connell, L., Convey, P., Fell, J.W., Frati, F., Hogg, I.D.,
Newsham, K.K., O’Donnell, A., Russell, N., Seppelt, R.D., Stevens, M.I., 2006.
Diversity and distribution of Victoria Land biota. Soil Biology & Biochemistry 38,
3003e3018.
Ball, B.A., Virginia, R.A., Barrett, J.E., Parsons, A.N., Wall, D.H., 2009. Interactions
between physical and biotic factors influence CO2 flux in Antarctic dry valley
soils. Soil Biology & Biochemistry 41, 1510e1517.
Barrett, P.J., Froggatt, P.C., 1978. Densities, porosities, and seismic velocities of some
rocks from Victoria Land, Antarctica. N.Z. Journal of Geology and Geophysics 21,
175e187.
Barrett, J.E., Virginia, R.A., Parsons, A.N., Wall, D.H., 2005. Potential soil organic
matter turnover in Taylor Valley, Antarctica. Arctic, Antarctic, and Alpine
Research 37, 108e117.
Barrett, J.E., Virginia, R.A., Parsons, A.N., Wall, D.H., 2006a. Soil carbon turnover in
the McMurdo Dry Valleys, Antarctica. Soil Biology & Biochemistry 38,
3065e3082.
Barrett, J.E., Virginia, R.A., Wall, D.H., Cary, S.C., Adams, B.J., Hacker, A.L.,
Aislabie, J.M., 2006b. Co-variation in soil biodiversity and biogeochemistry in
northern and southern Victoria Land, Antarctica. Antarctic Science 18, 535e548.
doi:10.1017/s0954102006000587.
Barrett, J.E., Virginia, R.A., Hopkins, D.W., Aislabie, J., Bargagli, R., Bockheim, J.G.,
Campbell, I.B., Lyons, W.B., Moorhead, D.L., Nkem, J.N., Sletten, R.S., Steltzer, H.,
Wall, D.H., Wallenstein, M.D., 2006c. Terrestrial ecosystem processes of Victoria
Land, Antarctica. Soil Biology & Biochemistry 38, 3019e3034.
F.L. Shanhun et al. / Soil Biology & Biochemistry 53 (2012) 99e111
Bockheim, J.G., Prentice, M.L., McLeod, M., 2008. Distribution of glacial deposits,
soils, and permafrost in Taylor Valley, Antarctica. Arctic, Antarctic, and Alpine
Research 40, 279e286. doi:10.1657/1523-0430(06-057).
Burkins, M.B., Virginia, R.A., Wall, D.H., 2001. Organic carbon cycling in Taylor Valley,
Antarctica: quantifying soil reservoirs and soil respiration. Global Change
Biology 7, 113e125.
Cary, S.C., McDonald, I.R., Barrett, J.E., Cowan, D.A., 2010. On the rocks: the
microbiology of Antarctic Dry Valley soils. Nature Reviews Microbiology 8,
129e138.
Doran, P.T., McKay, C.P., Clow, G.D., Dana, G.L., Fountain, A.G., Nylen, T., Lyons, W.B.,
2002. Valley floor climate observations from the McMurdo dry valleys, Antarctica,
1986e2000. Journal of Geophysical Research-Atmospheres 107.
Dörr, H., Münnich, K.O., 1980. Carbon-14 and carbon-13 in soil CO2. Radiocarbon 22,
909e918.
Elberling, B., Gregorich, E.G., Hopkins, D.W., Sparrow, A.D., Novis, P., Greenfield, L.G.,
2006. Distribution and dynamics of soil organic matter in an Antarctic dry
valley. Soil Biology & Biochemistry 38, 3095e3106.
Fang, C., Moncrieff, J.B., 2001. The dependence of soil CO2 efflux on temperature. Soil
Biology & Biochemistry 33, 155e165.
Fountain, A.G., Nylen, T.H., Monaghan, A., Basagic, H.J., Bromwich, D., 2010. Snow in
the McMurdo Dry Valleys, Antarctica. International Journal of Climatology 30,
633e642. doi:10.1002/joc.1933.
Fry, B., 2006. Stable Isotope Ecology. Springer, New York.
Gregorich, E.G., Hopkins, D.W., Elberling, B., Sparrow, A.D., Novis, P.,
Greenfield, L.G., Rochette, P., 2006. Emission of CO2, CH4 and N2O from
lakeshore soils in an Antarctic dry valley. Soil Biology & Biochemistry 38,
3120e3129.
Hall, B.L., Denton, G.H., 2000. Radiocarbon Chronology of Ross Sea Drift, Eastern
Taylor Valley, Antarctica: Evidence for a Grounded Ice Sheet in the Ross Sea at
the Last Glacial Maximum. Geografiska Annaler: Series A, Physical Geography
82, 305e336. doi:10.1111/j.0435-3676.2000.00127.x.
Harris, D., Horwath, W.R., van Kessel, C., 2001. Acid fumigation of soils to remove
carbonates prior to total organic carbon or carbon-13 isotopic analysis. Soil
Science Society of America Journal 65, 1853e1856.
Higgins, S.M., Hendy, C.H., Denton, G.H., 2000. Geochronology of Bonney drift,
Taylor Valley, Antarctica: evidence for Interglacial Expansions of Taylor Glacier.
Geografiska Annaler: Series A, Physical Geography 82, 391e409. doi:10.1111/
j.0435-3676.2000.00130.x.
Hopkins, D.W., Sparrow, A.D., Elberling, B., Gregorich, E.G., Novis, P.M.,
Greenfield, L.G., Tilston, E.L., 2006. Carbon, nitrogen and temperature controls
on microbial activity in soils from an Antarctic dry valley. Soil Biology &
Biochemistry 38, 3130e3140.
111
Hopkins, D.W., Sparrow, A.D., Shillam, L.L., English, L.C., Dennis, P.G., Novis, P.,
Elberling, B., Gregorich, E.G., Greenfield, L.G., 2008. Enzymatic activities and
microbial communities in an Antarctic Dry Valley soil: responses to C and N
supplementation. Soil Biology & Biochemistry 40, 2130e2136.
Hopkins, D.W., Sparrow, A.D., Gregorich, E.G., Elberling, B., Novis, P., Fraser, F.,
Scrimgeour, C., Dennis, P.G., Meier-Augenstein, W., Greenfield, L.G., 2009.
Isotopic evidence for the provenance and turnover of organic carbon by soil
microorganisms in the Antarctic Dry Valleys. Environmental Microbiology 11,
597e608. doi:10.1111/j.1462-2920.2008.01830.x.
Keeling, C.D., Piper, S.C., Bacastow, R.B., Wahlen, M., Whorf, T.P., Heimann, M.,
Meijer, H.A., 2001. Exchanges of Atmospheric CO2 and 13CO2 with the Terrestrial
Biosphere and Oceans from 1978 to 2000. I. Global aspects, SIO Reference
Series, No. 01-06. Scripps Institution of Oceanography, San Diego, 88 pp. http://
scrippsco82.ucsd.edu/data/atmospheric_co82.html.
Knohl, A., Werner, R.A., Geilmann, H., Brand, W.A., 2004. Kel-FÔ discs improve
storage time of canopy air samples in 10-mL vials for CO2-d13C analysis. Rapid
Communications in Mass Spectrometry 18, 1663e1665.
Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration.
Functional Ecology 8, 315e323.
Mook, W.G., Bommerson, J.C., Staverman, W.H., 1974. Carbon isotope
fractionation between dissolved bicarbonate and gaseous carbon dioxide.
Earth and Planetary Science Letters 22, 169e176. doi:10.1016/0012-821x(74)
90078-8.
Mook, W.G. (Ed.), 2000. Environmental isotopes in the hydrological cycle: Principles
and applications, vol. 1. UNESCO, Paris, p. 280.
Park, R., Epstein, S., 1960. Carbon isotope fractionation during photosynthesis.
Geochimica et Cosmochimica Acta 21, 110e126.
Parsons, A.N., Barrett, J.E., Wall, D.H., Virginia, R.A., 2004. Soil carbon dioxide flux in
Antarctic Dry Valley ecosystems. Ecosystems 7, 286e295.
Rochette, P., Bertrand, N., 2003. Soil air sample storage and handling using
polypropylene syringes and glass vials. Canadian Journal of Soil Science, 631e637.
Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry: Chemical Equilibria and Rates in
Natural Waters, third ed. Wiley Interscience, New York.
Thode, H.G., Shima, M., Rees, C.F., Krishnamurty, K.V., 1965. Carbon-13 isotope
effects in systems containing carbon dioxide, bicarbonate, carbonate and metal
ions. Canadian Journal of Chemistry 43, 582e595.
Treonis, A.M., Wall, D.H., Virginia, R.A., 2002. Field and microcosm studies of
decomposition and soil biota in a cold desert soil. Ecosystems 5, 159e170.
United States Salinity Laboratory Staff, 1954. Diagnosis and Improvement of Saline
and Alkali Soils. United States Department of Agriculture, Washington, D.C.
Vogel, J.C., Grootes, P.M., Mook, W.G., 1970. Isotope fractionation between gaseous
and dissolved carbon dioxide. Zeitschrift für Physik 230, 225e238.
Soil Biology & Biochemistry xxx (2013) 1e2
Contents lists available at ScienceDirect
Soil Biology & Biochemistry
journal homepage: www.elsevier.com/locate/soilbio
Comments on “Abiotic processes dominate CO2 fluxes in Antarctic soils” by
Shanhun et al. Soil Biology & Biochemistry 53, 99e111 (2012)
We appreciate the new insights into soil CO2 efflux in an Antarctic Dry Valley provided by Shanhun et al. (2012) in their recent paper. Their observations that abiotic processes may contribute to CO2
efflux and evidence for a dynamic equilibrium between CO2 in the
gaseous and dissolved phases which is influenced by temperature
and moisture content provide a mechanistic explanation for the
CO2 dynamics observed by Parsons et al. (2004) at low temperatures. However, to extrapolate from their observations which
were made during the warmest and coldest periods of two days
at two sites in one of the Dry Valleys (the Taylor Valley) in
Antarctica to all or any soil in Antarctica, as is implied by the title,
and to use these data to conclude that abiotic processes dominate
CO2 efflux is unsound.
As Shanhun et al. (2012) state, the Antarctic Dry Valleys represent the largest contiguous, or near contiguous, ice-free region of
Antarctica. However, the soils are far from typical of all soils in
Antarctica. The Dry Valley soils typically contain only traces of
organic matter, have very low water contents and remain frozen
for long periods of the year. This is in marked contrast to the ornithogenic soils in coastal regions and the maritime Antarctic, the
volcanic soils in Antarctica, the organic rich, peats at some sites
on the Antarctic Peninsula and on some islands, or the lithomorphic
soils in the mountains. Even within the Dry Valleys region, there is
significant diversity in soils (Bockheim et al., 2008) and associated
CO2 fluxes (Elberling et al., 2006) which are overlooked by Shanhun
et al. (2012). Put simply, failure to recognise the diversity of soils in
Antarctica at different scales limits the relevance of the observations: one model of CO2 cannot be applied to all Antarctic soils as
implied by Shanhun et al. (2012). Their blanket references to “the
Dry Valleys environment” mask the considerable pedological,
geological, biological, hydrological, and climatic heterogeneity
that must be accounted for in a large-scale model of soil CO2
exchange.
Shanhun et al. (2012) refer to our work in the Dry Valleys region
(Elberling et al., 2006; Hopkins et al., 2006a, 2006b, 2008, 2009;
Gregorich et al., 2006) and claim that their work challenges our biological interpretations of soil CO2 fluxes. However, very little of our
work is directly relevant to the Taylor Valley and only one paper
(Hopkins et al., 2009) contains any data for soils from the Taylor
Valley. In fact, much of our work, including more recently published
research (Sparrow et al., 2011; Dennis et al., 2013) has focused on
biological hotspots, such as near streams and at lake margins, and
on the effect of amendment of the soils with organic substrates
on soil microbial activity. There is compelling evidence to suggest
DOI of original article: http://dx.doi.org/10.1016/j.soilbio.2013.10.039.
that the response observed in subsidy experiments is biological in
origin. The response to added substrate applied in the field
occurred relatively rapidly and with a wide range of substrates.
As well, short-term incubation respiration rates measured under
controlled laboratory conditions (Hopkins et al., 2006a, 2006b;
Sparrow et al., 2011) are not influenced by these abiotic factors.
Hence before revising our view that CO2 efflux from Dry Valley soils
is predominantly biological, a mechanism is needed to explain the
increased CO2 efflux in subsidy experiments according to the process suggested by Shanhun et al. (2012).
Shanhun et al. (2012) developed a model of abiotic CO2 dynamics based on steady-state conditions at contrasting temperatures. However, under field conditions the processes of CO2
dissolution and degassing are kinetically limited and this limitation
increases at colder temperatures. Their measurements on warm
and cold days were conducted over a period of less than 2 h, but
it is well-established from pH measurements and groundwater
monitoring studies that it often takes up to 24 h for equilibrium between atmospheric and soil CO2 to be reached (Elberling and
Jakobsen, 2000).
The effluxes measured by Elberling et al. (2006) ranged from
0.02 to 0.05 mmol CO2eC m 2 s 1 for six landscape elements which
cover the majority of the land area in a Dry Valley (the Garwood
Valley), but rising as high as 0.80 mmol CO2eC m 2 s 1 at biological
hotspots. Shanhun et al.’s (2012) largest net efflux
(0.048 mmol CO2eC m 2 s 1) is in the same range. If, as Shanhun
et al. (2012) suggest, most of the CO2 efflux is driven by abiotic process, it is difficult to explain the measurements made by Elberling
et al. (2006) which showed a strong trend in soil CO2 efflux across
the landscape and correlated with soil organic carbon contents.
Furthermore, the d13C values for CO2 collected from dry valleys soils
incubated under stable temperature and moisture conditions
correspond closely to the d13C values for potential organic substrates for soil microorganisms, such as algal and cyanobacteria
detritus and dead lichen and moss biomass (Hopkins et al., 2009),
strongly supporting a biological mechanism for CO2 production.
We do not dispute that there could be an abiotic contribution to
CO2 efflux under some conditions (this was reported in 2004 by
Parsons et al.) and agree that determining these conditions and
the magnitude of their impact over a variety of landscape elements
is worthwhile. However, the evidence in Shanhun et al. (2012) is
not strong enough to dismiss biological processes dominating
CO2 efflux. More importantly, they are not strong enough to undermine one of the main conclusions of our work, which is that
contemporary sources of organic carbon are sufficient to sustain
current soil biological activity (Elberling et al., 2006; Hopkins
et al., 2009).
0038-0717/$ e see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soilbio.2013.10.046
Please cite this article in press as: Elberling, B., et al., Comments on “Abiotic processes dominate CO2 fluxes in Antarctic soils” by Shanhun et al.
Soil Biology & Biochemistry 53, 99e111 (2012), Soil Biology & Biochemistry (2013), http://dx.doi.org/10.1016/j.soilbio.2013.10.046
2
B. Elberling et al. / Soil Biology & Biochemistry xxx (2013) 1e2
In summary, therefore, we urge caution in extrapolating from
two sites in one of the Dry Valleys to all soils in Antarctica. The evidence of Shanhun et al. (2012) is not strong enough to support
their conclusion that “biological respiration makes only a minor
contribution to soil CO2 fluxes in the Dry Valleys environment”
based on measurements of “soil CO2 dynamics at two sites studied
in the Taylor Valley”.
Isotopic evidence for the provenance and turnover of organic carbon by soil microorganisms in the Antarctic Dry Valleys. Environ. Microbiol. 11, 597e608.
Parsons, A.N., Barrett, J.E., Wall, D.H., Virginia, R.A., 2004. Soil carbon dioxide flux in
Antarctic Dry Valley ecosystems. Ecosystems 7, 286e295.
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, M.S., 2012. Abiotic processes dominate CO2 fluxes in Antarctic soils. Soil Biol. Biochem. 53, 99e111. http://
dx.doi.org/10.1016/j.soilbio.2012.04.027.
Sparrow, A.D., Gregorich, E.G., Hopkins, D.W., Novis, P., Elberling, B., Greenfield, L.G.,
2011. Resource limitations on the activity of a soil microbial community in the
dry valleys of southern Victoria Land, Antarctica. Soil Sci. Soc. Am. J. 75, 2188e
2197.
References
Bockheim, J.G., Prentice, M.L., McLeod, M., 2008. Distribution of glacial deposits,
soils, and permafrost in Taylor Valley, Antarctica. Arctic Antarct. Alp. Res. 40,
279e286.
Dennis, P.G., Sparrow, A.D., Gregorich, E.G., Novis, P.M., Elberling, B., Greenfield, L.G.,
Hopkins, D.W., 2013. Microbial responses to carbon and nitrogen supplementation in an Antarctic Dry Valley soil. Antarct. Sci. 25, 55e61.
Elberling, B., Jakobsen, B.H., 2000. Soil solution pH measurements using in-line
chambers with tension lysimeters. Can. J. Soil Sci. 80, 283e288.
Elberling, B., Gregorich, E.G., Hopkins, D.W., Sparrow, A.D., Novis, P., Greenfield, L.G.,
2006. Distribution and dynamics of soil organic matter in an Antarctic Dry Valley. Soil Biol. Biochem. 38, 3095e3106.
Gregorich, E.G., Hopkins, D.W., Elberling, B., Novis, P., Greenfield, L.G., Sparrow, A.D.,
Rochette, P., 2006. Biogenic gas emission along a lake shore in an Antarctic Dry
Valley. Soil Biol. Biochem. 38, 3120e3129.
Hopkins, D.W., Sparrow, A.D., Elberling, B., Gregorich, E.G., Novis, P., Greenfield, L.G.,
Tilston, E.L., 2006a. Carbon, nitrogen and temperature controls on microbial activity in soils from an Antarctic Dry Valley. Soil Biol. Biochem. 38, 3130e3140.
Hopkins, D.W., Sparrow, A.D., Novis, P.M., Gregorich, E.G., Elberling, B.,
Greenfield, L.G., 2006b. Controls on the distribution of productivity and organic
resources in Antarctic Dry Valley soils. Proc. R. Soc. London B Biol. Sci. 273,
2687e2695.
Hopkins, D.W., Sparrow, A.D., Shillam, L.L., English, L.C., Dennis, P.G., Novis, P.M.,
Elberling, B., Gregorich, E.G., Greenfield, L.G., 2008. Enzymatic activities in Antarctic Dry Valley soil: responses to C and N supplementation. Soil Biol. Biochem.
40, 2130e2136.
Hopkins, D.W., Sparrow, A.D., Gregorich, E.G., Elberling, B., Novis, P., Fraser, F.,
Scrimgeour, C., Dennis, P.G., Meier-Augenstein, W., Greenfield, L.G., 2009.
B. Elberling
University of Copenhagen, Denmark
L.G. Greenfield
University of Canterbury, Christchurch, New Zealand
E.G. Gregorich*
Agriculture & Agri-Food Canada, Ottawa, Canada
D.W. Hopkins
Heriot-Watt University, Edinburgh, UK
P. Novis
Landcare Research, Lincoln, New Zealand
A.D. Sparrow
CSIRO, Alice Springs, Australia
* Corresponding author.
E-mail address: Ed.Gregorich@agr.gc.ca (E.G. Gregorich).
8 July 2013
Available online xxx
Please cite this article in press as: Elberling, B., et al., Comments on “Abiotic processes dominate CO2 fluxes in Antarctic soils” by Shanhun et al.
Soil Biology & Biochemistry 53, 99e111 (2012), Soil Biology & Biochemistry (2013), http://dx.doi.org/10.1016/j.soilbio.2013.10.046
Soil Biology & Biochemistry xxx (2013) 1e2
Contents lists available at ScienceDirect
Soil Biology & Biochemistry
journal homepage: www.elsevier.com/locate/soilbio
Letter to the Editor
Reply to Elberling et al.’s (2013) comments on “Abiotic processes dominate CO2
fluxes in Antarctic soils” (Soil Biol. Biochem. 53, 99e111)
We are pleased that Elberling et al. (2013) recognise that our
mechanistic model of abiotic CO2 dynamics represents a new
insight into understanding CO2 fluxes in Antarctic dry valley soils.
However, Elberling et al. (2013) disagree with our findings on two
fronts: first, they argue that the soils we studied are not representative of all Antarctic soils, and second, they question whether or
not the mechanism we have invoked is significant.
With regard to the first point, we acknowledge that our title,
used for the sake of brevity, could be misconstrued by those unfamiliar with the Antarctic ice-free areas. We studied two soils on
glacial drifts of different age in Taylor Valley, close to and distant
from contemporary organic carbon sources. We freely acknowledge
that these soils are not representative of ornithogenic soils, or soils
formed in peat, or where there are vascular plants on the Antarctic
Peninsula. The soils we studied are typical of the Typic Haploturbels,
and soil maps of the Taylor (Bockheim et al., 2008) and Wright Valleys (McLeod et al., 2009), along with other areas within the
McMurdo Dry Valleys region (Bockheim and McLeod, 2008), show
that Typic Haploturbels are a significant component of the soil landscape. Elberling et al. (2013) emphasise biological hotspots as representing the important characteristics of soil CO2 fluxes in Antarctica.
By contrast, our emphasis is placed on understanding soil CO2 fluxes
over the most extensive terrestrial ecosystems in continental
Antarctica. However, one of our sites, within 50 m of a lake, certainly
satisfies “hotspot” criteria. As such, we maintain that the results of
Shanhun et al. (2012) are relevant to a broader understanding of
CO2 dynamics in Antarctic terrestrial ecosystems, despite being
located within part of one of the McMurdo Dry Valleys.
The significance of the abiotic CO2 exsolutionedissolution
mechanism we invoked to explain CO2 dynamics is then downplayed by Elberling et al. (2013) on account of an inferred kinetic
limitation to the process. An example of this limitation was cited
from evidence of lags in pH changes in soil solution as a result of
temperature changes (Elberling and Jakobsen, 2000). The soils we
studied had <8% moisture, which, owing to their gravelly and
sandy nature, would correspond to a matric potential of at most
1000 kPa and probably less than 1500 kPa. In this circumstance,
water is held in very fine pores, or, given their likely absence, as thin
films on clasts. These very high surface area to volume ratio water
films would not suffer the same kinetic limitation as approximately
1 L of soil solution in a 2 L vessel with no soil matrix, as was the
experimental setup of Elberling and Jakobsen (2000).
Additional support for an abiotic mechanism of CO2 fluxes is
demonstrated via negative CO2 flux rates. The work of Parsons
DOI of original article: http://dx.doi.org/10.1016/j.soilbio.2012.04.027.
et al. (2004) and Ball et al. (2009) has shown negative CO2 fluxes
in soils of Taylor Valley, and our work confirms that these are driven
by diel variation in subsoil CO2 concentration gradients. These
fluxes have no biological explanation, and they have a similar
magnitude to the positive fluxes often attributed to heterotrophic
respiration. If these negative abiotic fluxes are not converted to positive fluxes on a diel cycle, then the soil must be storing carbon in an
inorganic form. To totally dismiss an abiotic efflux would require
that the soil stores about 220 mmol CaCO3/m2/day according to
our data. Assuming this diel variation happens over 60 days per
year (the number of days during which soil temperatures oscillate
above 0 C, hence liquid water is present), and that the process has
been going on for about 20,000 years (the age of the Ross Sea drift),
then the soil should be about 5% CaCO3 by mass. Our analyses show
it is 0.4% at the most.
Furthermore, recent work by Ma et al. (2013) clearly demonstrates through a field, laboratory and modelling study that the
same abiotic mechanism of CO2 fluxes that we invoke for Dry
Valley soils contributes significantly to the total soil CO2 flux
for saline/alkaline soils in an arid temperate environment. In
addition, Ma et al.’s (2013) data show that the abioticallydriven soil CO2 flux is even more dominant when soils are dry,
and that the effect is enhanced with both increasing pH and electrical conductivity. This suggests that our results are applicable to
an even larger proportion of Dry Valley soils than previously
considered, with soils classified as Typic Anhyorthels and Typic
Anhyturbels also likely to exhibit this effect. Together with the
broad applicability of this mechanism to Typic Haploturbels
(and Typic Haplorthels), and individual site characteristics
notwithstanding, this extends the area over which abiotic CO2
fluxes are likely to make a significant contribution to the total
soil CO2 flux to w93% of soils in the McMurdo Dry Valleys
(Bockheim and McLeod, 2008).
Finally, Elberling et al. (2013) “do not dispute that there could be
an abiotic contribution to CO2 efflux under some conditions”.
Consequently, Elberling et al.’s (2013) apprehension that our findings would “undermine one of the main conclusions of our [their]
work” is puzzling. Their conclusion “that contemporary sources of
organic carbon are sufficient to sustain current soil biological activity” is not challenged by our results. By identifying an important
abiotic contribution to soil CO2 fluxes, net heterotrophic respiration
rates would be lower than those previously inferred, hence shifting
the balance further towards a self-sustaining carbon economy.
In summary, Shanhun et al. (2012) present a novel,
mechanistically-based approach to understanding CO2 dynamics
in soils typical of large parts of the McMurdo Dry Valleys. We
0038-0717/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soilbio.2013.10.039
Please cite this article in press as: Shanhun, F.L., et al., Reply to Elberling et al.’s (2013) comments on “Abiotic processes dominate CO2 fluxes in
Antarctic soils” (Soil Biol. Biochem. 53, 99e111), Soil Biology & Biochemistry (2013), http://dx.doi.org/10.1016/j.soilbio.2013.10.039
2
Letter to the Editor / Soil Biology & Biochemistry xxx (2013) 1e2
maintain that our findings place an important caveat on interpretation of Dry Valley soil CO2 fluxes e they cannot be assumed to be
solely biologically driven.
References
Ball, B.A., Virginia, R.A., Barrett, J.E., Parsons, A.N., Wall, D.H., 2009. Interactions between physical and biotic factors influence CO2 flux in Antarctic dry valley soils.
Soil Biol. Biochem. 41, 1510e1517.
Bockheim, J.G., McLeod, M., 2008. Soil distribution in the McMurdo Dry Valleys,
Antarctica.
Geoderma
144,
43e49.
http://dx.doi.org/10.1016/
j.geoderma.2007.10.015.
Bockheim, J.G., Prentice, M.L., McLeod, M., 2008. Distribution of glacial deposits,
soils, and permafrost in Taylor Valley, Antarctica. Arct. Antarct. Alpine Res. 40,
279e286. http://dx.doi.org/10.1657/1523-0430(06-057).
Elberling, B., Jakobsen, B.H., 2000. Soil solution pH measurements using in-line
chambers with tension lysimeters. Can. J. Soil Sci. 80, 283e288.
Elberling, B., Greenfield, L.G., Gregorich, E.G., Hopkins, D.W., Novis, P., Sparrow, A.D.,
2013. Comments on “Abiotic processes dominate CO2 flux in Antarctic soils” by
Shanhun et al. Soil Biology & Biochemistry 53, 99e111 (2012). Soil Biol. Biochem. (Submitted for publication).
Ma, J., Wang, Z.Y., Stevenson, B.A., Zheng, X.J., Li, Y., 2013. An inorganic CO2 diffusion
and dissolution process explains negative CO2 fluxes in saline/alkaline soils. Sci.
Rep. 3, 2025. http://dx.doi.org/10.1038/srep02025.
McLeod, M., Bockheim, J., Balks, M., Aislabie, J., 2009. Soils of western Wright Valley,
Antarctica.
Antarct.
Sci.
21,
355e365.
http://dx.doi.org/10.1017/
S0954102009001965.
Parsons, A.N., Barrett, J.E., Wall, D.H., Virginia, R.A., 2004. Soil carbon dioxide flux in
Antarctic Dry Valley ecosystems. Ecosystems 7, 286e295.
Shanhun, F.L., Almond, P.C., Clough, T.J., Smith, C.M.S., 2012. Abiotic processes dominate CO2 fluxes in Antarctic soils. Soil Biol. Biochem. 53, 99e111. http://
dx.doi.org/10.1016/j.soilbio.2012.04.027.
Fiona L. Shanhun*, Peter C. Almond, Tim J. Clough
Department of Soil and Physical Sciences, PO Box 85084,
Lincoln University, Lincoln 7647, New Zealand
* Corresponding
author. Tel.: þ64 21 408 909; fax: þ64 3 325 3607.
E-mail addresses: fiona.shanhun@lincoln.ac.nz (F.L. Shanhun),
peter.almond@lincoln.ac.nz (P.C. Almond),
timothy.clough@lincoln.ac.nz (T.J. Clough).
18 October 2013
Available online xxx
Please cite this article in press as: Shanhun, F.L., et al., Reply to Elberling et al.’s (2013) comments on “Abiotic processes dominate CO2 fluxes in
Antarctic soils” (Soil Biol. Biochem. 53, 99e111), Soil Biology & Biochemistry (2013), http://dx.doi.org/10.1016/j.soilbio.2013.10.039